---
title: Stop reasons and fallback
url: https://platform.claude.com/docs/en/build-with-claude/handling-stop-reasons
description: Learn what each stop_reason value means and how to handle truncation, tool use, paused turns, and refusals in your application.
---
Every Messages API response includes a `stop_reason` field that tells you why Claude stopped generating. Check this field to decide whether to use the response as-is, continue the conversation, retry, or fall back to another model.
For the full response schema, see the [Messages API reference](https://platform.claude.com/docs/en/api/messages/create).
## Quick reference
| Value | When it occurs | What to do |
| -------------------------------------------------------------------------------------------------------------------------------------------- | ----------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------- |
| [`end_turn`](https://platform.claude.com/docs/en/build-with-claude/handling-stop-reasons#end-turn) | Claude finished its response naturally. | Use the response. |
| [`max_tokens`](https://platform.claude.com/docs/en/build-with-claude/handling-stop-reasons#max-tokens) | The response reached your `max_tokens` limit. | Raise `max_tokens` or [continue the response](https://platform.claude.com/docs/en/build-with-claude/handling-stop-reasons#ensuring-complete-responses). |
| [`stop_sequence`](https://platform.claude.com/docs/en/build-with-claude/handling-stop-reasons#stop-sequence) | Claude emitted one of your `stop_sequences`. | Read `stop_sequence` to see which one fired. |
| [`tool_use`](https://platform.claude.com/docs/en/build-with-claude/handling-stop-reasons#tool-use) | Claude is calling a tool. | Run the tool and return the result. A server tool call still missing its result block completes in a later response. |
| [`pause_turn`](https://platform.claude.com/docs/en/build-with-claude/handling-stop-reasons#pause-turn) | A server-tool loop reached its iteration limit. | Send the assistant content back to continue. |
| [`refusal`](https://platform.claude.com/docs/en/build-with-claude/handling-stop-reasons#refusal) | Claude declined to respond. | Read `stop_details` and [retry on a fallback model](https://platform.claude.com/docs/en/build-with-claude/refusals-and-fallback). |
| [`model_context_window_exceeded`](https://platform.claude.com/docs/en/build-with-claude/handling-stop-reasons#model-context-window-exceeded) | The response filled the model's context window. | Treat the response as truncated. |
## The stop\_reason field
The `stop_reason` field is part of every successful Messages API response. Unlike errors, which indicate failures in processing your request, `stop_reason` tells you why Claude completed its response generation.
```json Example response
{
"id": "msg_01234",
"type": "message",
"role": "assistant",
"content": [
{
"type": "text",
"text": "Here's the answer to your question..."
}
],
"stop_reason": "end_turn",
"stop_sequence": null,
"stop_details": null,
"usage": {
"input_tokens": 100,
"output_tokens": 50
}
}
```
## Stop reason values
### end\_turn
The most common stop reason. Indicates Claude finished its response naturally.
```bash cURL
curl https://api.anthropic.com/v1/messages \
-H "x-api-key: $ANTHROPIC_API_KEY" \
-H "anthropic-version: 2023-06-01" \
-H "content-type: application/json" \
-d '{
"model": "claude-opus-5-5",
"max_tokens": 1024,
"messages": [{"role": "user", "content": "Hello!"}]
}' | jq 'if .stop_reason == "end_turn" then (.content[] | select(.type == "text") | .text) else . end'
```
```bash CLI
ant messages create \
--model claude-opus-5-5 \
--max-tokens 1024 \
--message '{role: user, content: "Hello!"}' \
--format json | jq 'if .stop_reason == "end_turn" then (.content[] | select(.type == "text") | .text) else . end'
```
```python Python
client = anthropic.Anthropic()
response = client.messages.create(
model="claude-opus-5-5",
max_tokens=1024,
messages=[{"role": "user", "content": "Hello!"}],
)
if response.stop_reason == "end_turn":
# Process the complete response
for block in response.content:
if block.type == "text":
print(block.text)
```
```typescript TypeScript
const client = new Anthropic();
const response = await client.messages.create({
model: "claude-opus-5-5",
max_tokens: 1024,
messages: [{ role: "user", content: "Hello!" }]
});
if (response.stop_reason === "end_turn") {
// Process the complete response
const textBlock = response.content.find(
(block): block is Anthropic.TextBlock => block.type === "text"
);
console.log(textBlock?.text);
}
```
```csharp C#
AnthropicClient client = new();
var response = await client.Messages.Create(new MessageCreateParams
{
Model = Model.ClaudeOpus5_5,
MaxTokens = 1024,
Messages = [new() { Role = Role.User, Content = "Hello!" }]
});
if (response.StopReason == "end_turn")
{
// Process the complete response
foreach (var block in response.Content)
{
if (block.TryPickText(out var textBlock))
{
Console.WriteLine(textBlock.Text);
}
}
}
```
```go Go
client := anthropic.NewClient()
response, err := client.Messages.New(context.TODO(), anthropic.MessageNewParams{
Model: anthropic.ModelClaudeOpus5_5,
MaxTokens: 1024,
Messages: []anthropic.MessageParam{
anthropic.NewUserMessage(anthropic.NewTextBlock("Hello!")),
},
})
if err != nil {
log.Fatal(err)
}
if response.StopReason == "end_turn" {
// Process the complete response
for _, block := range response.Content {
if textBlock, ok := block.AsAny().(anthropic.TextBlock); ok {
fmt.Println(textBlock.Text)
}
}
}
```
```java Java
AnthropicClient client = AnthropicOkHttpClient.fromEnv();
Message response = client.messages().create(
MessageCreateParams.builder()
.model(Model.CLAUDE_OPUS_5_5)
.maxTokens(1024L)
.addUserMessage("Hello!")
.build()
);
if (response.stopReason().map(StopReason.END_TURN::equals).orElse(false)) {
// Process the complete response
response.content().stream()
.flatMap(block -> block.text().stream())
.forEach(textBlock -> IO.println(textBlock.text()));
}
```
```php PHP
$client = new Client();
$response = $client->messages->create(
maxTokens: 1024,
messages: [['role' => 'user', 'content' => 'Hello!']],
model: 'claude-opus-5-5',
);
if ($response->stopReason === 'end_turn') {
// Process the complete response
foreach ($response->content as $block) {
if ($block->type === 'text') {
echo $block->text, PHP_EOL;
}
}
}
```
```ruby Ruby
client = Anthropic::Client.new
response = client.messages.create(
model: "claude-opus-5-5",
max_tokens: 1024,
messages: [{ role: "user", content: "Hello!" }]
)
if response.stop_reason == :end_turn
# Process the complete response
response.content.each do |block|
puts block.text if block.type == :text
end
end
```
Sometimes Claude returns an empty response (exactly 2–3 tokens with no content) with `stop_reason: "end_turn"`. This typically occurs when Claude interprets that the assistant turn is complete, particularly after tool results.
**Common causes:**
* Adding text blocks immediately after tool results (Claude learns to expect the user to always insert text after tool results, so it ends its turn to follow the pattern)
* Sending Claude's completed response back without adding anything (Claude already determined it's done, so it will remain done)
**How to prevent empty responses:**
```python Python
# INCORRECT: Adding text immediately after tool_result
messages = [
{"role": "user", "content": "Calculate the sum of 1234 and 5678"},
{
"role": "assistant",
"content": [
{
"type": "tool_use",
"id": "toolu_123",
"name": "calculator",
"input": {"operation": "add", "a": 1234, "b": 5678},
}
],
},
{
"role": "user",
"content": [
{"type": "tool_result", "tool_use_id": "toolu_123", "content": "6912"},
{
"type": "text",
"text": "Here's the result", # Don't add text after tool_result
},
],
},
]
# CORRECT: Send tool results directly without additional text
messages = [
{"role": "user", "content": "Calculate the sum of 1234 and 5678"},
{
"role": "assistant",
"content": [
{
"type": "tool_use",
"id": "toolu_123",
"name": "calculator",
"input": {"operation": "add", "a": 1234, "b": 5678},
}
],
},
{
"role": "user",
"content": [
{"type": "tool_result", "tool_use_id": "toolu_123", "content": "6912"}
],
}, # Just the tool_result, no additional text
]
```
```typescript TypeScript
// INCORRECT: Adding text immediately after tool_result
let messages: Anthropic.MessageParam[] = [
{ role: "user", content: "Calculate the sum of 1234 and 5678" },
{
role: "assistant",
content: [
{
type: "tool_use",
id: "toolu_123",
name: "calculator",
input: { operation: "add", a: 1234, b: 5678 }
}
]
},
{
role: "user",
content: [
{ type: "tool_result", tool_use_id: "toolu_123", content: "6912" },
{ type: "text", text: "Here's the result" } // Don't add text after tool_result
]
}
];
// CORRECT: Send tool results directly without additional text
messages = [
{ role: "user", content: "Calculate the sum of 1234 and 5678" },
{
role: "assistant",
content: [
{
type: "tool_use",
id: "toolu_123",
name: "calculator",
input: { operation: "add", a: 1234, b: 5678 }
}
]
},
{
role: "user",
// Just the tool_result, no additional text
content: [{ type: "tool_result", tool_use_id: "toolu_123", content: "6912" }]
}
];
```
```csharp C#
using System.Text.Json;
using Anthropic.Models.Messages;
var input = JsonSerializer.Deserialize>(
"""{"operation":"add","a":1234,"b":5678}"""
)!;
// INCORRECT: Adding text immediately after tool_result
List messages =
[
new() { Role = Role.User, Content = "Calculate the sum of 1234 and 5678" },
new()
{
Role = Role.Assistant,
Content = new List
{
new ToolUseBlockParam { ID = "toolu_123", Name = "calculator", Input = input }
}
},
new()
{
Role = Role.User,
Content = new List
{
new ToolResultBlockParam { ToolUseID = "toolu_123", Content = "6912" },
new TextBlockParam { Text = "Here's the result" } // Don't add text after tool_result
}
}
];
// CORRECT: Send tool results directly without additional text
messages =
[
new() { Role = Role.User, Content = "Calculate the sum of 1234 and 5678" },
new()
{
Role = Role.Assistant,
Content = new List
{
new ToolUseBlockParam { ID = "toolu_123", Name = "calculator", Input = input }
}
},
new()
{
Role = Role.User,
// Just the tool_result, no additional text
Content = new List
{
new ToolResultBlockParam { ToolUseID = "toolu_123", Content = "6912" }
}
}
];
```
```go Go
input := map[string]any{"operation": "add", "a": 1234, "b": 5678}
// INCORRECT: Adding text immediately after tool_result
messages := []anthropic.MessageParam{
anthropic.NewUserMessage(anthropic.NewTextBlock("Calculate the sum of 1234 and 5678")),
anthropic.NewAssistantMessage(
anthropic.NewToolUseBlock("toolu_123", input, "calculator"),
),
anthropic.NewUserMessage(
anthropic.NewToolResultBlock("toolu_123", "6912", false),
anthropic.NewTextBlock("Here's the result"), // Don't add text after tool_result
),
}
// CORRECT: Send tool results directly without additional text
messages = []anthropic.MessageParam{
anthropic.NewUserMessage(anthropic.NewTextBlock("Calculate the sum of 1234 and 5678")),
anthropic.NewAssistantMessage(
anthropic.NewToolUseBlock("toolu_123", input, "calculator"),
),
// Just the tool_result, no additional text
anthropic.NewUserMessage(
anthropic.NewToolResultBlock("toolu_123", "6912", false),
),
}
```
```java Java
ToolUseBlockParam toolUse = ToolUseBlockParam.builder()
.id("toolu_123")
.name("calculator")
.input(ToolUseBlockParam.Input.builder()
.putAdditionalProperty("operation", JsonValue.from("add"))
.putAdditionalProperty("a", JsonValue.from(1234))
.putAdditionalProperty("b", JsonValue.from(5678))
.build())
.build();
// INCORRECT: Adding text immediately after tool_result
List messages = List.of(
MessageParam.builder().role(MessageParam.Role.USER)
.content("Calculate the sum of 1234 and 5678").build(),
MessageParam.builder().role(MessageParam.Role.ASSISTANT)
.contentOfBlockParams(List.of(ContentBlockParam.ofToolUse(toolUse))).build(),
MessageParam.builder().role(MessageParam.Role.USER)
.contentOfBlockParams(List.of(
ContentBlockParam.ofToolResult(
ToolResultBlockParam.builder().toolUseId("toolu_123").content("6912").build()),
// Don't add text after tool_result
ContentBlockParam.ofText(TextBlockParam.builder().text("Here's the result").build())
)).build()
);
// CORRECT: Send tool results directly without additional text
messages = List.of(
MessageParam.builder().role(MessageParam.Role.USER)
.content("Calculate the sum of 1234 and 5678").build(),
MessageParam.builder().role(MessageParam.Role.ASSISTANT)
.contentOfBlockParams(List.of(ContentBlockParam.ofToolUse(toolUse))).build(),
// Just the tool_result, no additional text
MessageParam.builder().role(MessageParam.Role.USER)
.contentOfBlockParams(List.of(
ContentBlockParam.ofToolResult(
ToolResultBlockParam.builder().toolUseId("toolu_123").content("6912").build())
)).build()
);
```
```php PHP
// INCORRECT: Adding text immediately after tool_result
$messages = [
['role' => 'user', 'content' => 'Calculate the sum of 1234 and 5678'],
[
'role' => 'assistant',
'content' => [
[
'type' => 'tool_use',
'id' => 'toolu_123',
'name' => 'calculator',
'input' => ['operation' => 'add', 'a' => 1234, 'b' => 5678],
],
],
],
[
'role' => 'user',
'content' => [
['type' => 'tool_result', 'tool_use_id' => 'toolu_123', 'content' => '6912'],
// Don't add text after tool_result
['type' => 'text', 'text' => "Here's the result"],
],
],
];
// CORRECT: Send tool results directly without additional text
$messages = [
['role' => 'user', 'content' => 'Calculate the sum of 1234 and 5678'],
[
'role' => 'assistant',
'content' => [
[
'type' => 'tool_use',
'id' => 'toolu_123',
'name' => 'calculator',
'input' => ['operation' => 'add', 'a' => 1234, 'b' => 5678],
],
],
],
[
'role' => 'user',
// Just the tool_result, no additional text
'content' => [
['type' => 'tool_result', 'tool_use_id' => 'toolu_123', 'content' => '6912'],
],
],
];
```
```ruby Ruby
# INCORRECT: Adding text immediately after tool_result
messages = [
{ role: "user", content: "Calculate the sum of 1234 and 5678" },
{
role: "assistant",
content: [
{
type: "tool_use",
id: "toolu_123",
name: "calculator",
input: { operation: "add", a: 1234, b: 5678 }
}
]
},
{
role: "user",
content: [
{ type: "tool_result", tool_use_id: "toolu_123", content: "6912" },
# Don't add text after tool_result
{ type: "text", text: "Here's the result" }
]
}
]
# CORRECT: Send tool results directly without additional text
messages = [
{ role: "user", content: "Calculate the sum of 1234 and 5678" },
{
role: "assistant",
content: [
{
type: "tool_use",
id: "toolu_123",
name: "calculator",
input: { operation: "add", a: 1234, b: 5678 }
}
]
},
{
role: "user",
# Just the tool_result, no additional text
content: [
{ type: "tool_result", tool_use_id: "toolu_123", content: "6912" }
]
}
]
```
If you still get empty responses after fixing the message structure, add a continuation prompt in a new user message rather than retrying with the empty response:
```python Python
def handle_empty_response(client, messages):
response = client.messages.create(
model="claude-opus-5-5", max_tokens=1024, messages=messages
)
# Check if response is empty
if response.stop_reason == "end_turn" and not response.content:
# INCORRECT: Don't just retry with the empty response
# This won't work because Claude already decided it's done
# CORRECT: Add a continuation prompt in a NEW user message
messages.append({"role": "user", "content": "Please continue"})
response = client.messages.create(
model="claude-opus-5-5", max_tokens=1024, messages=messages
)
return response
```
```typescript TypeScript
async function handleEmptyResponse(
client: Anthropic,
messages: Anthropic.MessageParam[]
): Promise {
let response = await client.messages.create({
model: "claude-opus-5-5",
max_tokens: 1024,
messages
});
// Check if response is empty
if (response.stop_reason === "end_turn" && response.content.length === 0) {
// INCORRECT: Don't just retry with the empty response
// This won't work because Claude already decided it's done
// CORRECT: Add a continuation prompt in a NEW user message
messages.push({ role: "user", content: "Please continue" });
response = await client.messages.create({
model: "claude-opus-5-5",
max_tokens: 1024,
messages
});
}
return response;
}
```
```csharp C#
static async Task HandleEmptyResponse(AnthropicClient client, List messages)
{
var response = await client.Messages.Create(new MessageCreateParams
{
Model = Model.ClaudeOpus5_5,
MaxTokens = 1024,
Messages = messages
});
// Check if response is empty
if (response.StopReason == "end_turn" && response.Content.Count == 0)
{
// CORRECT: Add a continuation prompt in a NEW user message
messages.Add(new() { Role = Role.User, Content = "Please continue" });
response = await client.Messages.Create(new MessageCreateParams
{
Model = Model.ClaudeOpus5_5,
MaxTokens = 1024,
Messages = messages
});
}
return response;
}
```
```go Go
func handleEmptyResponse(client anthropic.Client, messages []anthropic.MessageParam) (*anthropic.Message, error) {
response, err := client.Messages.New(context.TODO(), anthropic.MessageNewParams{
Model: anthropic.ModelClaudeOpus5_5,
MaxTokens: 1024,
Messages: messages,
})
if err != nil {
return nil, err
}
// Check if response is empty
if response.StopReason == "end_turn" && len(response.Content) == 0 {
// CORRECT: Add a continuation prompt in a NEW user message
messages = append(messages, anthropic.NewUserMessage(anthropic.NewTextBlock("Please continue")))
response, err = client.Messages.New(context.TODO(), anthropic.MessageNewParams{
Model: anthropic.ModelClaudeOpus5_5,
MaxTokens: 1024,
Messages: messages,
})
if err != nil {
return nil, err
}
}
return response, nil
}
```
```java Java
static Message handleEmptyResponse(AnthropicClient client, List messages) {
Message response = client.messages().create(
MessageCreateParams.builder()
.model(Model.CLAUDE_OPUS_5_5)
.maxTokens(1024L)
.messages(messages)
.build()
);
// Check if response is empty
boolean isEndTurn = response.stopReason().map(StopReason.END_TURN::equals).orElse(false);
if (isEndTurn && response.content().isEmpty()) {
// CORRECT: Add a continuation prompt in a NEW user message
List extended = new ArrayList<>(messages);
extended.add(MessageParam.builder()
.role(MessageParam.Role.USER)
.content("Please continue")
.build());
response = client.messages().create(
MessageCreateParams.builder()
.model(Model.CLAUDE_OPUS_5_5)
.maxTokens(1024L)
.messages(extended)
.build()
);
}
return response;
}
```
```php PHP
function handle_empty_response(Client $client, array $messages)
{
$response = $client->messages->create(
maxTokens: 1024,
messages: $messages,
model: 'claude-opus-5-5',
);
// Check if response is empty
if ($response->stopReason === 'end_turn' && count($response->content) === 0) {
// CORRECT: Add a continuation prompt in a NEW user message
$messages[] = ['role' => 'user', 'content' => 'Please continue'];
$response = $client->messages->create(
maxTokens: 1024,
messages: $messages,
model: 'claude-opus-5-5',
);
}
return $response;
}
```
```ruby Ruby
def handle_empty_response(client, messages)
response = client.messages.create(
model: "claude-opus-5-5",
max_tokens: 1024,
messages: messages
)
# Check if response is empty
if response.stop_reason == :end_turn && response.content.empty?
# CORRECT: Add a continuation prompt in a NEW user message
messages << { role: "user", content: "Please continue" }
response = client.messages.create(
model: "claude-opus-5-5",
max_tokens: 1024,
messages: messages
)
end
response
end
```
**Best practices:**
1. **Never add text blocks immediately after tool results:** This teaches Claude to expect user input after every tool use.
2. **Don't retry empty responses without modification:** Sending the empty response back won't help.
3. **Use continuation prompts as a last resort:** Only if these fixes don't resolve the issue.
### max\_tokens
Claude stopped because it reached the `max_tokens` limit specified in your request.
```bash cURL
curl https://api.anthropic.com/v1/messages \
-H "x-api-key: $ANTHROPIC_API_KEY" \
-H "anthropic-version: 2023-06-01" \
-H "content-type: application/json" \
-d '{
"model": "claude-opus-5-5",
"max_tokens": 10,
"messages": [{"role": "user", "content": "Explain quantum physics"}]
}' | jq '.stop_reason'
```
```bash CLI
ant messages create \
--model claude-opus-5-5 \
--max-tokens 10 \
--message '{role: user, content: "Explain quantum physics"}' \
--format json | jq '.stop_reason'
```
```python Python
client = anthropic.Anthropic()
# Request with limited tokens
response = client.messages.create(
model="claude-opus-5-5",
max_tokens=10,
messages=[{"role": "user", "content": "Explain quantum physics"}],
)
if response.stop_reason == "max_tokens":
# Response was truncated
print("Response was cut off at token limit")
# Consider making another request to continue
```
```typescript TypeScript
const client = new Anthropic();
// Request with limited tokens
const response = await client.messages.create({
model: "claude-opus-5-5",
max_tokens: 10,
messages: [{ role: "user", content: "Explain quantum physics" }]
});
if (response.stop_reason === "max_tokens") {
// Response was truncated
console.log("Response was cut off at token limit");
// Consider making another request to continue
}
```
```csharp C#
AnthropicClient client = new();
// Request with limited tokens
var response = await client.Messages.Create(new MessageCreateParams
{
Model = Model.ClaudeOpus5_5,
MaxTokens = 10,
Messages = [new() { Role = Role.User, Content = "Explain quantum physics" }]
});
if (response.StopReason == "max_tokens")
{
// Response was truncated
Console.WriteLine("Response was cut off at token limit");
// Consider making another request to continue
}
```
```go Go
client := anthropic.NewClient()
// Request with limited tokens
response, err := client.Messages.New(context.TODO(), anthropic.MessageNewParams{
Model: anthropic.ModelClaudeOpus5_5,
MaxTokens: 10,
Messages: []anthropic.MessageParam{
anthropic.NewUserMessage(anthropic.NewTextBlock("Explain quantum physics")),
},
})
if err != nil {
log.Fatal(err)
}
if response.StopReason == "max_tokens" {
// Response was truncated
fmt.Println("Response was cut off at token limit")
// Consider making another request to continue
}
```
```java Java
AnthropicClient client = AnthropicOkHttpClient.fromEnv();
// Request with limited tokens
Message response = client.messages().create(
MessageCreateParams.builder()
.model(Model.CLAUDE_OPUS_5_5)
.maxTokens(10L)
.addUserMessage("Explain quantum physics")
.build()
);
if (response.stopReason().map(StopReason.MAX_TOKENS::equals).orElse(false)) {
// Response was truncated
IO.println("Response was cut off at token limit");
// Consider making another request to continue
}
```
```php PHP
$client = new Client();
// Request with limited tokens
$response = $client->messages->create(
maxTokens: 10,
messages: [['role' => 'user', 'content' => 'Explain quantum physics']],
model: 'claude-opus-5-5',
);
if ($response->stopReason === 'max_tokens') {
// Response was truncated
echo 'Response was cut off at token limit', PHP_EOL;
// Consider making another request to continue
}
```
```ruby Ruby
client = Anthropic::Client.new
# Request with limited tokens
response = client.messages.create(
model: "claude-opus-5-5",
max_tokens: 10,
messages: [{ role: "user", content: "Explain quantum physics" }]
)
if response.stop_reason == :max_tokens
# Response was truncated
puts "Response was cut off at token limit"
# Consider making another request to continue
end
```
If Claude's response is cut off because it hit the `max_tokens` limit, and the truncated response contains an incomplete tool use block, you'll need to retry the request with a higher `max_tokens` value to get the full tool use.
```bash CLI
RESPONSE=$(ant messages create --max-tokens 1024 --format jsonl <<'YAML'
model: claude-opus-5-5
tools:
- name: get_weather
description: Get the current weather in a given location
input_schema:
type: object
properties:
location:
type: string
required:
- location
messages:
- role: user
content: What is the weather in San Francisco?
YAML
)
# Check if the response was truncated mid tool use
STOP_REASON=$(jq -r '.stop_reason' <<<"$RESPONSE")
LAST_TYPE=$(jq -r '.content[-1].type' <<<"$RESPONSE")
if [ "$STOP_REASON" = "max_tokens" ] && [ "$LAST_TYPE" = "tool_use" ]; then
# Retry with a higher max_tokens
ant messages create --max-tokens 4096 <<'YAML'
model: claude-opus-5-5
tools:
- name: get_weather
description: Get the current weather in a given location
input_schema:
type: object
properties:
location:
type: string
required:
- location
messages:
- role: user
content: What is the weather in San Francisco?
YAML
fi
```
```python Python
# Check if response was truncated during tool use
if response.stop_reason == "max_tokens":
# Check if the last content block is an incomplete tool_use
last_block = response.content[-1]
if last_block.type == "tool_use":
# Send the request with higher max_tokens
response = client.messages.create(
model="claude-opus-5-5",
max_tokens=4096, # Increased limit
messages=messages,
tools=tools,
)
```
```typescript TypeScript
// Check if response was truncated during tool use
if (response.stop_reason === "max_tokens") {
// Check if the last content block is an incomplete tool_use
const lastBlock = response.content[response.content.length - 1];
if (lastBlock.type === "tool_use") {
// Send the request with higher max_tokens
response = await client.messages.create({
model: "claude-opus-5-5",
max_tokens: 4096, // Increased limit
messages: messages,
tools: tools
});
}
}
```
```csharp C#
using System.Linq;
using Anthropic;
using Anthropic.Models.Messages;
AnthropicClient client = new();
var parameters = new MessageCreateParams
{
Model = Model.ClaudeOpus5_5,
MaxTokens = 1024,
Messages = messages,
Tools = tools
};
var response = await client.Messages.Create(parameters);
if (response.StopReason == "max_tokens")
{
var lastBlock = response.Content.Last();
if (lastBlock.TryPickToolUse(out _))
{
response = await client.Messages.Create(parameters with { MaxTokens = 4096 });
}
}
```
```go Go
response, err := client.Messages.New(context.TODO(), anthropic.MessageNewParams{
Model: anthropic.ModelClaudeOpus5_5,
MaxTokens: 1024,
Messages: messages,
Tools: tools,
})
if err != nil {
log.Fatal(err)
}
if response.StopReason == "max_tokens" {
lastBlock := response.Content[len(response.Content)-1]
switch lastBlock.AsAny().(type) {
case anthropic.ToolUseBlock:
response, err = client.Messages.New(context.TODO(), anthropic.MessageNewParams{
Model: anthropic.ModelClaudeOpus5_5,
MaxTokens: 4096,
Messages: messages,
Tools: tools,
})
if err != nil {
log.Fatal(err)
}
}
}
```
```java Java
// Check if response was truncated during tool use
if (response.stopReason().isPresent() && response.stopReason().get().equals(StopReason.MAX_TOKENS)) {
ContentBlock lastBlock = response.content().get(response.content().size() - 1);
if (lastBlock.toolUse().isPresent()) {
// Send the request with higher max_tokens
response = client.messages().create(
MessageCreateParams.builder()
.model(Model.CLAUDE_OPUS_5_5)
.maxTokens(4096L) // Increased limit
.messages(messages)
.tools(tools)
.build()
);
}
}
```
```php PHP
$response = $client->messages->create(
maxTokens: 1024,
messages: $messages,
model: 'claude-opus-5-5',
tools: $tools,
);
if ($response->stopReason === 'max_tokens') {
$lastBlock = end($response->content);
if ($lastBlock->type === 'tool_use') {
$response = $client->messages->create(
maxTokens: 4096,
messages: $messages,
model: 'claude-opus-5-5',
tools: $tools,
);
}
}
```
```ruby Ruby
response = client.messages.create(
model: "claude-opus-5-5",
max_tokens: 1024,
messages: messages,
tools: tools
)
if response.stop_reason == :max_tokens
last_block = response.content.last
if last_block.type == :tool_use
response = client.messages.create(
model: "claude-opus-5-5",
max_tokens: 4096,
messages: messages,
tools: tools
)
end
end
```
### stop\_sequence
Claude encountered one of your custom stop sequences.
```bash cURL
curl https://api.anthropic.com/v1/messages \
-H "x-api-key: $ANTHROPIC_API_KEY" \
-H "anthropic-version: 2023-06-01" \
-H "content-type: application/json" \
-d '{
"model": "claude-opus-5-5",
"max_tokens": 1024,
"stop_sequences": ["END", "STOP"],
"messages": [{"role": "user", "content": "Generate text until you say END"}]
}' | jq '{stop_reason, stop_sequence}'
```
```bash CLI
ant messages create \
--model claude-opus-5-5 \
--max-tokens 1024 \
--stop-sequence END --stop-sequence STOP \
--message '{role: user, content: "Generate text until you say END"}' \
--format json | jq '{stop_reason, stop_sequence}'
```
```python Python
client = anthropic.Anthropic()
response = client.messages.create(
model="claude-opus-5-5",
max_tokens=1024,
stop_sequences=["END", "STOP"],
messages=[{"role": "user", "content": "Generate text until you say END"}],
)
if response.stop_reason == "stop_sequence":
print(f"Stopped at sequence: {response.stop_sequence}")
```
```typescript TypeScript
const client = new Anthropic();
const response = await client.messages.create({
model: "claude-opus-5-5",
max_tokens: 1024,
stop_sequences: ["END", "STOP"],
messages: [{ role: "user", content: "Generate text until you say END" }]
});
if (response.stop_reason === "stop_sequence") {
console.log(`Stopped at sequence: ${response.stop_sequence}`);
}
```
```csharp C#
AnthropicClient client = new();
var response = await client.Messages.Create(new MessageCreateParams
{
Model = Model.ClaudeOpus5_5,
MaxTokens = 1024,
StopSequences = ["END", "STOP"],
Messages = [new() { Role = Role.User, Content = "Generate text until you say END" }]
});
if (response.StopReason == "stop_sequence")
{
Console.WriteLine($"Stopped at sequence: {response.StopSequence}");
}
```
```go Go
client := anthropic.NewClient()
response, err := client.Messages.New(context.TODO(), anthropic.MessageNewParams{
Model: anthropic.ModelClaudeOpus5_5,
MaxTokens: 1024,
StopSequences: []string{"END", "STOP"},
Messages: []anthropic.MessageParam{
anthropic.NewUserMessage(anthropic.NewTextBlock("Generate text until you say END")),
},
})
if err != nil {
log.Fatal(err)
}
if response.StopReason == "stop_sequence" {
fmt.Printf("Stopped at sequence: %s\n", response.StopSequence)
}
```
```java Java
AnthropicClient client = AnthropicOkHttpClient.fromEnv();
Message response = client.messages().create(
MessageCreateParams.builder()
.model(Model.CLAUDE_OPUS_5_5)
.maxTokens(1024L)
.addStopSequence("END")
.addStopSequence("STOP")
.addUserMessage("Generate text until you say END")
.build()
);
if (response.stopReason().map(StopReason.STOP_SEQUENCE::equals).orElse(false)) {
IO.println("Stopped at sequence: " + response.stopSequence().orElse(""));
}
```
```php PHP
$client = new Client();
$response = $client->messages->create(
maxTokens: 1024,
messages: [['role' => 'user', 'content' => 'Generate text until you say END']],
model: 'claude-opus-5-5',
stopSequences: ['END', 'STOP'],
);
if ($response->stopReason === 'stop_sequence') {
echo "Stopped at sequence: {$response->stopSequence}", PHP_EOL;
}
```
```ruby Ruby
client = Anthropic::Client.new
response = client.messages.create(
model: "claude-opus-5-5",
max_tokens: 1024,
stop_sequences: ["END", "STOP"],
messages: [{ role: "user", content: "Generate text until you say END" }]
)
if response.stop_reason == :stop_sequence
puts "Stopped at sequence: #{response.stop_sequence}"
end
```
### tool\_use
Claude is calling a tool and expects you to run it.
For most tool use implementations, use the [tool runner](https://platform.claude.com/docs/en/agents-and-tools/tool-use/tool-runner), which automatically handles tool execution, result formatting, and conversation management.
```bash cURL
curl https://api.anthropic.com/v1/messages \
-H "x-api-key: $ANTHROPIC_API_KEY" \
-H "anthropic-version: 2023-06-01" \
-H "content-type: application/json" \
-d '{
"model": "claude-opus-5-5",
"max_tokens": 1024,
"tools": [{
"name": "get_weather",
"description": "Get the current weather in a given location",
"input_schema": {
"type": "object",
"properties": {"location": {"type": "string", "description": "City and state"}},
"required": ["location"]
}
}],
"messages": [{"role": "user", "content": "What is the weather in San Francisco?"}]
}' | jq '.stop_reason, (.content[] | select(.type == "tool_use"))'
```
```bash CLI
ant messages create --format json <<'YAML' | jq '.stop_reason, (.content[] | select(.type == "tool_use"))'
model: claude-opus-5-5
max_tokens: 1024
messages:
- role: user
content: What is the weather in San Francisco?
tools:
- name: get_weather
description: Get the current weather in a given location
input_schema:
type: object
properties:
location: {type: string, description: City and state}
required: [location]
YAML
```
```python Python
client = anthropic.Anthropic()
weather_tool = {
"name": "get_weather",
"description": "Get the current weather in a given location",
"input_schema": {
"type": "object",
"properties": {
"location": {"type": "string", "description": "City and state"},
},
"required": ["location"],
},
}
def execute_tool(name, tool_input):
"""Execute a tool and return the result."""
return f"Weather in {tool_input.get('location', 'unknown')}: 72°F"
response = client.messages.create(
model="claude-opus-5-5",
max_tokens=1024,
tools=[weather_tool],
messages=[{"role": "user", "content": "What is the weather in San Francisco?"}],
)
if response.stop_reason == "tool_use":
# Extract and execute the tool
for block in response.content:
if block.type == "tool_use":
result = execute_tool(block.name, block.input)
# Return result to Claude for final response
```
```typescript TypeScript
const client = new Anthropic();
const weatherTool: Anthropic.Tool = {
name: "get_weather",
description: "Get the current weather in a given location",
input_schema: {
type: "object",
properties: {
location: { type: "string", description: "City and state" }
},
required: ["location"]
}
};
function executeTool(name: string, input: Record): string {
return `Weather in ${input.location ?? "unknown"}: 72°F`;
}
const response = await client.messages.create({
model: "claude-opus-5-5",
max_tokens: 1024,
tools: [weatherTool],
messages: [{ role: "user", content: "What is the weather in San Francisco?" }]
});
if (response.stop_reason === "tool_use") {
// Extract and execute the tool
for (const block of response.content) {
if (block.type === "tool_use") {
const result = executeTool(block.name, block.input as Record);
// Return result to Claude for final response
}
}
}
```
```csharp C#
AnthropicClient client = new();
var weatherTool = new Tool
{
Name = "get_weather",
Description = "Get the current weather in a given location",
InputSchema = new InputSchema
{
Properties = new Dictionary
{
["location"] = JsonSerializer.SerializeToElement(
new { type = "string", description = "City and state" }
),
},
Required = ["location"]
}
};
var response = await client.Messages.Create(new MessageCreateParams
{
Model = Model.ClaudeOpus5_5,
MaxTokens = 1024,
Tools = [weatherTool],
Messages = [new() { Role = Role.User, Content = "What is the weather in San Francisco?" }]
});
if (response.StopReason == "tool_use")
{
// Extract and execute the tool
foreach (var block in response.Content)
{
if (block.TryPickToolUse(out var toolUse))
{
// Execute toolUse.Name with toolUse.Input and return the result to Claude
}
}
}
```
```go Go
client := anthropic.NewClient()
weatherTool := anthropic.ToolParam{
Name: "get_weather",
Description: anthropic.String("Get the current weather in a given location"),
InputSchema: anthropic.ToolInputSchemaParam{
Properties: map[string]any{
"location": map[string]string{"type": "string", "description": "City and state"},
},
Required: []string{"location"},
},
}
response, err := client.Messages.New(context.TODO(), anthropic.MessageNewParams{
Model: anthropic.ModelClaudeOpus5_5,
MaxTokens: 1024,
Tools: []anthropic.ToolUnionParam{{OfTool: &weatherTool}},
Messages: []anthropic.MessageParam{
anthropic.NewUserMessage(anthropic.NewTextBlock("What is the weather in San Francisco?")),
},
})
if err != nil {
log.Fatal(err)
}
if response.StopReason == "tool_use" {
// Extract and execute the tool
for _, block := range response.Content {
if toolUse, ok := block.AsAny().(anthropic.ToolUseBlock); ok {
fmt.Println(toolUse.Name, toolUse.Input)
// Return result to Claude for final response
}
}
}
```
```java Java
void main() {
AnthropicClient client = AnthropicOkHttpClient.fromEnv();
Tool weatherTool = Tool.builder()
.name("get_weather")
.description("Get the current weather in a given location")
.inputSchema(Tool.InputSchema.builder()
.properties(JsonValue.from(Map.of(
"location", Map.of("type", "string", "description", "City and state")
)))
.putAdditionalProperty("required", JsonValue.from(List.of("location")))
.build())
.build();
Message response = client.messages().create(
MessageCreateParams.builder()
.model(Model.CLAUDE_OPUS_5_5)
.maxTokens(1024L)
.addTool(weatherTool)
.addUserMessage("What is the weather in San Francisco?")
.build()
);
if (response.stopReason().map(StopReason.TOOL_USE::equals).orElse(false)) {
// Extract and execute the tool
for (ContentBlock block : response.content()) {
block.toolUse().ifPresent(toolUse -> {
// Execute toolUse.name() with toolUse.input() and return the result to Claude
});
}
}
```
```php PHP
$client = new Client();
$weatherTool = [
'name' => 'get_weather',
'description' => 'Get the current weather in a given location',
'input_schema' => [
'type' => 'object',
'properties' => [
'location' => ['type' => 'string', 'description' => 'City and state'],
],
'required' => ['location'],
],
];
$response = $client->messages->create(
maxTokens: 1024,
messages: [['role' => 'user', 'content' => 'What is the weather in San Francisco?']],
model: 'claude-opus-5-5',
tools: [$weatherTool],
);
if ($response->stopReason === 'tool_use') {
// Extract and execute the tool
foreach ($response->content as $block) {
if ($block->type === 'tool_use') {
// Execute $block->name with $block->input and return the result to Claude
}
}
}
```
```ruby Ruby
client = Anthropic::Client.new
weather_tool = {
name: "get_weather",
description: "Get the current weather in a given location",
input_schema: {
type: "object",
properties: {
location: { type: "string", description: "City and state" }
},
required: ["location"]
}
}
response = client.messages.create(
model: "claude-opus-5-5",
max_tokens: 1024,
tools: [weather_tool],
messages: [{ role: "user", content: "What is the weather in San Francisco?" }]
)
if response.stop_reason == :tool_use
# Extract and execute the tool
response.content.each do |block|
next unless block.type == :tool_use
# Execute block.name with block.input and return the result to Claude
end
end
```
A `tool_use` response can also contain a `server_tool_use` block whose `id` has no matching result block. That server tool call is not finished, and this response does not carry its result. In the common case, Claude calls a [server tool](https://platform.claude.com/docs/en/agents-and-tools/tool-use/server-tools) and one of your client tools in the same group of parallel tool calls: the API returns without running the server tool so that you can run the client tools first. There is no other marker for the state; detect it by checking each `server_tool_use` or `mcp_tool_use` block's `id` for a matching result block.
With [programmatic tool calling](https://platform.claude.com/docs/en/agents-and-tools/tool-use/programmatic-tool-calling), the same response shape means something different. The client `tool_use` block comes from code that is running in the `code_execution` tool rather than from Claude directly, and its `caller` field names the `code_execution` block that called it. That code has already started: it is paused waiting for your `tool_result` blocks, and sending them resumes the execution instead of starting a deferred tool. The `code_execution` block's own result block arrives once the code finishes, which can take more than one round of tool results. The follow-up user message itself is the same in both cases; with programmatic tool calling, also pass back the `id` from the response's `container` field, as that page shows.
```json A mixed tool_use response
{
"stop_reason": "tool_use",
"content": [
{
"type": "server_tool_use",
"id": "srvtoolu_01HxbWnMRmbWyMfUtJKC45rA",
"name": "web_search",
"input": { "query": "example article" }
},
{
"type": "tool_use",
"id": "toolu_01PjgRJLbXrXEMZwDNYLnBqk",
"name": "run_command",
"input": { "command": "uname -a" }
}
]
}
```
The continuation is a user message of `tool_result` blocks, one for every `tool_use` block in the response (see [Handle tool calls](https://platform.claude.com/docs/en/agents-and-tools/tool-use/handle-tool-calls)), with two extra rules: that message must contain nothing except the `tool_result` blocks, and the request must keep the same `tools` array. A resume request that no longer defines the waiting server tool fails with a 400 whose message ends ``but no `web_search` tool was provided``. The API attaches your results to the still-open assistant turn, runs the deferred server tool (for paused code execution, resumes it), and continues the turn. For a server tool Claude called directly, the next response's `content` starts with the result block that answers the previous response's `server_tool_use` `id`.
```json The follow-up user message
{
"role": "user",
"content": [
{
"type": "tool_result",
"tool_use_id": "toolu_01PjgRJLbXrXEMZwDNYLnBqk",
"content": "Linux demo-host 6.8.0-52-generic x86_64 GNU/Linux"
}
]
}
```
Adding anything after the `tool_result` blocks in that user message, such as text, ends the assistant turn; for a server tool Claude called directly, the request then fails with a 400 `invalid_request_error` that names the unresolved server tool:
```text wrap
`web_search` tool use with id `srvtoolu_01HxbWnMRmbWyMfUtJKC45rA` was found without a corresponding `web_search_tool_result` block
```
Leaving out a `tool_result`, or putting one after other content, fails earlier with the standard `tool_use ids were found without tool_result blocks immediately after` error instead. To give Claude more input, send it as a separate user message after the turn completes.
### pause\_turn
Returned when the server-side sampling loop reaches its iteration limit while executing [server tools](https://platform.claude.com/docs/en/agents-and-tools/tool-use/server-tools) such as web search. The default limit is 10 iterations per request.
When this happens, the response may contain a `server_tool_use` block without a corresponding result block. To let Claude finish processing, continue the conversation by sending the response back as-is. A response that leaves a client `tool_use` block waiting on you never has a `stop_reason` of `pause_turn`: when Claude stops to call your tools, `stop_reason` is [`tool_use`](https://platform.claude.com/docs/en/build-with-claude/handling-stop-reasons#tool-use), and you continue it by sending the client `tool_result` blocks instead of the response itself.
```bash cURL
# The SDKs handle continuation directly. With cURL, inspect stop_reason
# on the response and re-POST with the assistant content appended.
curl https://api.anthropic.com/v1/messages \
-H "x-api-key: $ANTHROPIC_API_KEY" \
-H "anthropic-version: 2023-06-01" \
-H "content-type: application/json" \
-d '{
"model": "claude-opus-5-5",
"max_tokens": 4096,
"tools": [{"type": "web_search_20250305", "name": "web_search"}],
"messages": [{"role": "user", "content": "Search for latest AI news"}]
}' | jq '{stop_reason, content}'
```
```bash CLI
# Inspect stop_reason; if it is pause_turn, re-run with the assistant
# response appended to --message.
ant messages create --format json <<'YAML' | jq '{stop_reason, content}'
model: claude-opus-5-5
max_tokens: 4096
tools:
- {type: web_search_20250305, name: web_search}
messages:
- {role: user, content: "Search for latest AI news"}
YAML
```
```python Python
response = client.messages.create(
model="claude-opus-5-5",
max_tokens=4096,
tools=[{"type": "web_search_20250305", "name": "web_search"}],
messages=[{"role": "user", "content": "Search for latest AI news"}],
)
if response.stop_reason == "pause_turn":
# Continue the conversation by sending the response back
messages = [
{"role": "user", "content": "Search for latest AI news"},
{"role": "assistant", "content": response.content},
]
continuation = client.messages.create(
model="claude-opus-5-5",
max_tokens=4096,
messages=messages,
tools=[{"type": "web_search_20250305", "name": "web_search"}],
)
```
```typescript TypeScript
const response = await client.messages.create({
model: "claude-opus-5-5",
max_tokens: 4096,
tools: [{ type: "web_search_20250305", name: "web_search" }],
messages: [{ role: "user", content: "Search for latest AI news" }]
});
if (response.stop_reason === "pause_turn") {
// Continue the conversation by sending the response back
const continuation = await client.messages.create({
model: "claude-opus-5-5",
max_tokens: 4096,
tools: [{ type: "web_search_20250305", name: "web_search" }],
messages: [
{ role: "user", content: "Search for latest AI news" },
{ role: "assistant", content: response.content }
]
});
}
```
```csharp C#
List tools = [new ToolUnion(new WebSearchTool20250305())];
MessageParam userMessage = new() { Role = Role.User, Content = "Search for latest AI news" };
var response = await client.Messages.Create(new MessageCreateParams
{
Model = Model.ClaudeOpus5_5,
MaxTokens = 4096,
Tools = tools,
Messages = [userMessage]
});
if (response.StopReason == "pause_turn")
{
// Continue the conversation by sending the response back
var continuation = await client.Messages.Create(new MessageCreateParams
{
Model = Model.ClaudeOpus5_5,
MaxTokens = 4096,
Tools = tools,
Messages =
[
userMessage,
new()
{
Role = Role.Assistant,
Content = response.Content.Select(block => new ContentBlockParam(block.Json)).ToList()
}
]
});
}
```
```go Go
tools := []anthropic.ToolUnionParam{
{OfWebSearchTool20250305: &anthropic.WebSearchTool20250305Param{}},
}
userMessage := anthropic.NewUserMessage(anthropic.NewTextBlock("Search for latest AI news"))
response, err := client.Messages.New(context.TODO(), anthropic.MessageNewParams{
Model: anthropic.ModelClaudeOpus5_5,
MaxTokens: 4096,
Tools: tools,
Messages: []anthropic.MessageParam{userMessage},
})
if err != nil {
log.Fatal(err)
}
if response.StopReason == "pause_turn" {
// Continue the conversation by sending the response back
var contentParams []anthropic.ContentBlockParamUnion
for _, block := range response.Content {
contentParams = append(contentParams, block.ToParam())
}
continuation, err := client.Messages.New(context.TODO(), anthropic.MessageNewParams{
Model: anthropic.ModelClaudeOpus5_5,
MaxTokens: 4096,
Tools: tools,
Messages: []anthropic.MessageParam{userMessage, anthropic.NewAssistantMessage(contentParams...)},
})
if err != nil {
log.Fatal(err)
}
_ = continuation
}
```
```java Java
Message response = client.messages().create(
MessageCreateParams.builder()
.model(Model.CLAUDE_OPUS_5_5)
.maxTokens(4096L)
.addTool(WebSearchTool20250305.builder().build())
.addUserMessage("Search for latest AI news")
.build()
);
if (response.stopReason().map(StopReason.PAUSE_TURN::equals).orElse(false)) {
// Continue the conversation by sending the response back
Message continuation = client.messages().create(
MessageCreateParams.builder()
.model(Model.CLAUDE_OPUS_5_5)
.maxTokens(4096L)
.addTool(WebSearchTool20250305.builder().build())
.addUserMessage("Search for latest AI news")
.addMessage(response)
.build()
);
}
```
```php PHP
$tools = [['type' => 'web_search_20250305', 'name' => 'web_search']];
$userMessage = ['role' => 'user', 'content' => 'Search for latest AI news'];
$response = $client->messages->create(
maxTokens: 4096,
messages: [$userMessage],
model: 'claude-opus-5-5',
tools: $tools,
);
if ($response->stopReason === 'pause_turn') {
// Continue the conversation by sending the response back
$continuation = $client->messages->create(
maxTokens: 4096,
messages: [
$userMessage,
['role' => 'assistant', 'content' => $response->content],
],
model: 'claude-opus-5-5',
tools: $tools,
);
}
```
```ruby Ruby
tools = [{ type: "web_search_20250305", name: "web_search" }]
user_message = { role: "user", content: "Search for latest AI news" }
response = client.messages.create(
model: "claude-opus-5-5",
max_tokens: 4096,
tools: tools,
messages: [user_message]
)
if response.stop_reason == :pause_turn
# Continue the conversation by sending the response back
continuation = client.messages.create(
model: "claude-opus-5-5",
max_tokens: 4096,
tools: tools,
messages: [user_message, { role: "assistant", content: response.content }]
)
end
```
Your application should handle `pause_turn` in any agent loop that uses server tools. Add the assistant's response to your messages array and make another API request to let Claude continue.
### refusal
Claude declined to generate a response. Safety classifiers return this stop reason as a normal HTTP 200 response, not an error.
```bash cURL
curl https://api.anthropic.com/v1/messages \
-H "x-api-key: $ANTHROPIC_API_KEY" \
-H "anthropic-version: 2023-06-01" \
-H "content-type: application/json" \
-d '{
"model": "claude-opus-5-5",
"max_tokens": 1024,
"messages": [{"role": "user", "content": "[Unsafe request]"}]
}' | jq '{stop_reason, stop_details}'
```
```bash CLI
ant messages create \
--model claude-opus-5-5 \
--max-tokens 1024 \
--message '{role: user, content: "[Unsafe request]"}' \
--format json | jq '{stop_reason, stop_details}'
```
```python Python
client = anthropic.Anthropic()
response = client.messages.create(
model="claude-opus-5-5",
max_tokens=1024,
messages=[{"role": "user", "content": "[Unsafe request]"}],
)
if response.stop_reason == "refusal":
# Claude declined to respond
print("Claude was unable to process this request")
# Consider rephrasing or modifying the request
```
```typescript TypeScript
const client = new Anthropic();
const response = await client.messages.create({
model: "claude-opus-5-5",
max_tokens: 1024,
messages: [{ role: "user", content: "[Unsafe request]" }]
});
if (response.stop_reason === "refusal") {
// Claude declined to respond
console.log("Claude was unable to process this request");
// Consider rephrasing or modifying the request
}
```
```csharp C#
AnthropicClient client = new();
var response = await client.Messages.Create(new MessageCreateParams
{
Model = Model.ClaudeOpus5_5,
MaxTokens = 1024,
Messages = [new() { Role = Role.User, Content = "[Unsafe request]" }]
});
if (response.StopReason == "refusal")
{
// Claude declined to respond
Console.WriteLine("Claude was unable to process this request");
// Consider rephrasing or modifying the request
}
```
```go Go
client := anthropic.NewClient()
response, err := client.Messages.New(context.TODO(), anthropic.MessageNewParams{
Model: anthropic.ModelClaudeOpus5_5,
MaxTokens: 1024,
Messages: []anthropic.MessageParam{
anthropic.NewUserMessage(anthropic.NewTextBlock("[Unsafe request]")),
},
})
if err != nil {
log.Fatal(err)
}
if response.StopReason == "refusal" {
// Claude declined to respond
fmt.Println("Claude was unable to process this request")
// Consider rephrasing or modifying the request
}
```
```java Java
AnthropicClient client = AnthropicOkHttpClient.fromEnv();
Message response = client.messages().create(
MessageCreateParams.builder()
.model(Model.CLAUDE_OPUS_5_5)
.maxTokens(1024L)
.addUserMessage("[Unsafe request]")
.build()
);
if (response.stopReason().map(StopReason.REFUSAL::equals).orElse(false)) {
// Claude declined to respond
IO.println("Claude was unable to process this request");
// Consider rephrasing or modifying the request
}
```
```php PHP
$client = new Client();
$response = $client->messages->create(
maxTokens: 1024,
messages: [['role' => 'user', 'content' => '[Unsafe request]']],
model: 'claude-opus-5-5',
);
if ($response->stopReason === 'refusal') {
// Claude declined to respond
echo 'Claude was unable to process this request', PHP_EOL;
// Consider rephrasing or modifying the request
}
```
```ruby Ruby
client = Anthropic::Client.new
response = client.messages.create(
model: "claude-opus-5-5",
max_tokens: 1024,
messages: [{ role: "user", content: "[Unsafe request]" }]
)
if response.stop_reason == :refusal
# Claude declined to respond
puts "Claude was unable to process this request"
# Consider rephrasing or modifying the request
end
```
If you encounter `refusal` stop reasons frequently while using Claude Sonnet 4.5 or Claude Opus 4.1 (the latter [retired, except on Bedrock and Google Cloud](https://platform.claude.com/docs/en/about-claude/model-deprecations)), you can try updating your API calls to use Haiku 4.5 (`claude-haiku-4-5-20251001`), which has different usage restrictions. Learn more about [understanding Sonnet 4.5's API safety filters](https://support.claude.com/en/articles/12449294-understanding-sonnet-4-5-s-api-safety-filters).
On a refusal, the `stop_details` object identifies the policy category that triggered it. The categories and the full refusal response shape are covered on [Refusals and fallback](https://platform.claude.com/docs/en/build-with-claude/refusals-and-fallback#refusal-response). `stop_details` is `null` for all stop reasons other than `refusal`.
A refused request on Claude Fable 5.1, Claude Fable 5, Claude Opus 5.5, Claude Opus 5, or Claude Sonnet 5.5 can usually be served by retrying on another Claude model. [Refusals and fallback](https://platform.claude.com/docs/en/build-with-claude/refusals-and-fallback) shows how to set up that retry, server-side or in your client. If you build the retry yourself from Claude Fable 5.1, Claude Fable 5, Claude Opus 5.5, Claude Opus 5, or Claude Sonnet 5.5, [fallback credit](https://platform.claude.com/docs/en/build-with-claude/fallback-credit) covers how to avoid paying the prompt-cache cost twice.
### model\_context\_window\_exceeded
Claude stopped because it reached the model's context window limit. This lets you request the maximum possible tokens without knowing the exact input size.
This stop reason is currently typed only in the SDK's `beta` (csharp, go: `Beta`; java: `beta()`) namespace, so the following examples call `client.beta.messages` (csharp, go: `client.Beta.Messages`; java: `client.beta().messages()`; php: `$client->beta->messages`) and use the `Beta`-prefixed types. On Sonnet 4.5 and newer models the API returns this value without a beta header. For earlier models, add the `model-context-window-exceeded-2025-08-26` beta header to enable it.
```bash cURL
curl https://api.anthropic.com/v1/messages \
-H "x-api-key: $ANTHROPIC_API_KEY" \
-H "anthropic-version: 2023-06-01" \
-H "content-type: application/json" \
-d '{
"model": "claude-opus-5-5",
"max_tokens": 20000,
"messages": [{"role": "user", "content": "Large input that uses most of context window..."}]
}' | jq '.stop_reason'
```
```bash CLI
ant messages create \
--model claude-opus-5-5 \
--max-tokens 20000 \
--message '{role: user, content: "Large input that uses most of context window..."}' \
--format json | jq '.stop_reason'
```
```python Python
# Request with maximum tokens to get as much as possible
response = client.beta.messages.create(
model="claude-opus-5-5",
max_tokens=20000, # Python SDK requires streaming for max_tokens above ~21k
messages=[
{"role": "user", "content": "Large input that uses most of context window..."}
],
)
if response.stop_reason == "model_context_window_exceeded":
# Response hit context window limit before max_tokens
print("Response reached model's context window limit")
# The response is still valid but was limited by context window
```
```typescript TypeScript
// Request with maximum tokens to get as much as possible
const response = await client.beta.messages.create({
model: "claude-opus-5-5",
max_tokens: 20000,
messages: [{ role: "user", content: "Large input that uses most of context window..." }]
});
if (response.stop_reason === "model_context_window_exceeded") {
// Response hit context window limit before max_tokens
console.log("Response reached model's context window limit");
// The response is still valid but was limited by context window
}
```
```csharp C#
using Anthropic.Models.Beta.Messages;
using Model = Anthropic.Models.Messages.Model;
// Request with maximum tokens to get as much as possible
var response = await client.Beta.Messages.Create(new MessageCreateParams
{
Model = Model.ClaudeOpus5_5,
MaxTokens = 20000,
Messages = [new() { Role = Role.User, Content = "Large input that uses most of context window..." }]
});
if (response.StopReason?.Value() == BetaStopReason.ModelContextWindowExceeded)
{
// Response hit context window limit before max_tokens
Console.WriteLine("Response reached model's context window limit");
// The response is still valid but was limited by context window
}
```
```go Go
// Request with maximum tokens to get as much as possible
response, err := client.Beta.Messages.New(context.TODO(), anthropic.BetaMessageNewParams{
Model: anthropic.ModelClaudeOpus5_5,
MaxTokens: 20000,
Messages: []anthropic.BetaMessageParam{
anthropic.NewBetaUserMessage(anthropic.NewBetaTextBlock("Large input that uses most of context window...")),
},
})
if err != nil {
log.Fatal(err)
}
if response.StopReason == anthropic.BetaStopReasonModelContextWindowExceeded {
// Response hit context window limit before max_tokens
fmt.Println("Response reached model's context window limit")
// The response is still valid but was limited by context window
}
```
```java Java
import com.anthropic.models.beta.messages.BetaMessage;
import com.anthropic.models.beta.messages.BetaStopReason;
import com.anthropic.models.beta.messages.MessageCreateParams;
// Request with maximum tokens to get as much as possible
BetaMessage response = client.beta().messages().create(
MessageCreateParams.builder()
.model(Model.CLAUDE_OPUS_5_5)
.maxTokens(20000L)
.addUserMessage("Large input that uses most of context window...")
.build()
);
if (response.stopReason().map(BetaStopReason.MODEL_CONTEXT_WINDOW_EXCEEDED::equals).orElse(false)) {
// Response hit context window limit before max_tokens
IO.println("Response reached model's context window limit");
// The response is still valid but was limited by context window
}
```
```php PHP
// Request with maximum tokens to get as much as possible
$response = $client->beta->messages->create(
maxTokens: 20000,
messages: [['role' => 'user', 'content' => 'Large input that uses most of context window...']],
model: 'claude-opus-5-5',
);
if ($response->stopReason === 'model_context_window_exceeded') {
// Response hit context window limit before max_tokens
echo 'Response reached model\'s context window limit', PHP_EOL;
// The response is still valid but was limited by context window
}
```
```ruby Ruby
# Request with maximum tokens to get as much as possible
response = client.beta.messages.create(
model: "claude-opus-5-5",
max_tokens: 20000,
messages: [{ role: "user", content: "Large input that uses most of context window..." }]
)
if response.stop_reason == :model_context_window_exceeded
# Response hit context window limit before max_tokens
puts "Response reached model's context window limit"
# The response is still valid but was limited by context window
end
```
## Best practices for handling stop reasons
### Always check stop\_reason
Make it a habit to check the `stop_reason` in your response handling logic:
```python Python
def handle_response(response):
match response.stop_reason:
case "tool_use":
return handle_tool_use(response)
case "max_tokens":
return handle_truncation(response)
case "model_context_window_exceeded":
return handle_context_limit(response)
case "pause_turn":
return handle_pause(response)
case "refusal":
return handle_refusal(response)
case _:
# Handle end_turn and other cases
return next(
(block.text for block in response.content if block.type == "text"),
"",
)
```
```typescript TypeScript
function handleResponse(response: Anthropic.Beta.BetaMessage): string {
switch (response.stop_reason) {
case "tool_use":
return handleToolUse(response);
case "max_tokens":
return handleTruncation(response);
case "model_context_window_exceeded":
return handleContextLimit(response);
case "pause_turn":
return handlePause(response);
case "refusal":
return handleRefusal(response);
default: {
// Handle end_turn and other cases
const textBlock = response.content.find(
(block): block is Anthropic.Beta.BetaTextBlock => block.type === "text"
);
return textBlock?.text ?? "";
}
}
}
```
```csharp C#
static string HandleResponse(BetaMessage response)
{
return response.StopReason?.Value() switch
{
BetaStopReason.ToolUse => HandleToolUse(response),
BetaStopReason.MaxTokens => HandleTruncation(response),
BetaStopReason.ModelContextWindowExceeded => HandleContextLimit(response),
BetaStopReason.PauseTurn => HandlePause(response),
BetaStopReason.Refusal => HandleRefusal(response),
// Handle end_turn and other cases
_ => response.Content.Select(b => b.Value).OfType().FirstOrDefault()?.Text ?? "",
};
}
```
```go Go
func handleResponse(response *anthropic.BetaMessage) string {
switch response.StopReason {
case anthropic.BetaStopReasonToolUse:
return handleToolUse(response)
case anthropic.BetaStopReasonMaxTokens:
return handleTruncation(response)
case anthropic.BetaStopReasonModelContextWindowExceeded:
return handleContextLimit(response)
case anthropic.BetaStopReasonPauseTurn:
return handlePause(response)
case anthropic.BetaStopReasonRefusal:
return handleRefusal(response)
default:
// Handle end_turn and other cases
for _, block := range response.Content {
if textBlock, ok := block.AsAny().(anthropic.BetaTextBlock); ok {
return textBlock.Text
}
}
return ""
}
}
```
```java Java
static String handleResponse(BetaMessage response) {
BetaStopReason reason = response.stopReason().orElse(BetaStopReason.END_TURN);
if (reason.equals(BetaStopReason.TOOL_USE)) {
return handleToolUse(response);
} else if (reason.equals(BetaStopReason.MAX_TOKENS)) {
return handleTruncation(response);
} else if (reason.equals(BetaStopReason.MODEL_CONTEXT_WINDOW_EXCEEDED)) {
return handleContextLimit(response);
} else if (reason.equals(BetaStopReason.PAUSE_TURN)) {
return handlePause(response);
} else if (reason.equals(BetaStopReason.REFUSAL)) {
return handleRefusal(response);
}
// Handle end_turn and other cases
return response.content().stream()
.filter(BetaContentBlock::isText)
.findFirst()
.map(block -> block.asText().text())
.orElse("");
}
```
```php PHP
function handle_response($response): string
{
return match ($response->stopReason) {
'tool_use' => handle_tool_use($response),
'max_tokens' => handle_truncation($response),
'model_context_window_exceeded' => handle_context_limit($response),
'pause_turn' => handle_pause($response),
'refusal' => handle_refusal($response),
// Handle end_turn and other cases
default => array_find($response->content, static fn ($block): bool => $block->type === 'text')?->text ?? '',
};
}
```
```ruby Ruby
def handle_response(response)
case response.stop_reason
when :tool_use then handle_tool_use(response)
when :max_tokens then handle_truncation(response)
when :model_context_window_exceeded then handle_context_limit(response)
when :pause_turn then handle_pause(response)
when :refusal then handle_refusal(response)
else
# Handle end_turn and other cases
response.content.find { it.type == :text }&.text
end
end
```
### Handle truncated responses gracefully
When a response is truncated because of token limits or the context window, append a notice so the reader knows the output is incomplete. To continue generating from where the response left off instead, see [Ensuring complete responses](https://platform.claude.com/docs/en/build-with-claude/handling-stop-reasons#ensuring-complete-responses).
```python Python
def handle_truncated_response(response):
text = next((block.text for block in response.content if block.type == "text"), "")
if response.stop_reason in ["max_tokens", "model_context_window_exceeded"]:
if response.stop_reason == "max_tokens":
note = "[Response truncated due to max_tokens limit]"
else:
note = "[Response truncated due to context window limit]"
return f"{text}\n\n{note}"
return text
```
```typescript TypeScript
function handleTruncatedResponse(response: Anthropic.Beta.BetaMessage): string {
const textBlock = response.content.find(
(block): block is Anthropic.Beta.BetaTextBlock => block.type === "text"
);
const text = textBlock?.text ?? "";
if (
response.stop_reason === "max_tokens" ||
response.stop_reason === "model_context_window_exceeded"
) {
const note =
response.stop_reason === "max_tokens"
? "[Response truncated due to max_tokens limit]"
: "[Response truncated due to context window limit]";
return `${text}\n\n${note}`;
}
return text;
}
```
```csharp C#
static string HandleTruncatedResponse(BetaMessage response)
{
var text = response.Content.Select(b => b.Value).OfType().FirstOrDefault()?.Text ?? "";
var reason = response.StopReason?.Value();
if (reason is BetaStopReason.MaxTokens or BetaStopReason.ModelContextWindowExceeded)
{
var note = reason == BetaStopReason.MaxTokens
? "[Response truncated due to max_tokens limit]"
: "[Response truncated due to context window limit]";
return $"{text}\n\n{note}";
}
return text;
}
```
```go Go
func handleTruncatedResponse(response *anthropic.BetaMessage) string {
text := ""
for _, block := range response.Content {
if textBlock, ok := block.AsAny().(anthropic.BetaTextBlock); ok {
text = textBlock.Text
break
}
}
if response.StopReason == anthropic.BetaStopReasonMaxTokens ||
response.StopReason == anthropic.BetaStopReasonModelContextWindowExceeded {
note := "[Response truncated due to context window limit]"
if response.StopReason == anthropic.BetaStopReasonMaxTokens {
note = "[Response truncated due to max_tokens limit]"
}
return text + "\n\n" + note
}
return text
}
```
```java Java
static String handleTruncatedResponse(BetaMessage response) {
String text = response.content().stream()
.filter(BetaContentBlock::isText)
.findFirst()
.map(block -> block.asText().text())
.orElse("");
BetaStopReason reason = response.stopReason().orElse(BetaStopReason.END_TURN);
if (reason.equals(BetaStopReason.MAX_TOKENS)
|| reason.equals(BetaStopReason.MODEL_CONTEXT_WINDOW_EXCEEDED)) {
String note = reason.equals(BetaStopReason.MAX_TOKENS)
? "[Response truncated due to max_tokens limit]"
: "[Response truncated due to context window limit]";
return text + "\n\n" + note;
}
return text;
}
```
```php PHP
function handle_truncated_response($response): string
{
$text = array_find($response->content, static fn ($block): bool => $block->type === 'text')?->text ?? '';
if (in_array($response->stopReason, ['max_tokens', 'model_context_window_exceeded'], true)) {
$note = $response->stopReason === 'max_tokens'
? '[Response truncated due to max_tokens limit]'
: '[Response truncated due to context window limit]';
return "{$text}\n\n{$note}";
}
return $text;
}
```
```ruby Ruby
def handle_truncated_response(response)
text = response.content.find { it.type == :text }&.text
if [:max_tokens, :model_context_window_exceeded].include?(response.stop_reason)
note = if response.stop_reason == :max_tokens
"[Response truncated due to max_tokens limit]"
else
"[Response truncated due to context window limit]"
end
return "#{text}\n\n#{note}"
end
text
end
```
### Implement retry logic for pause\_turn
When using [server tools](https://platform.claude.com/docs/en/agents-and-tools/tool-use/server-tools), the API may return `pause_turn` if the server-side sampling loop reaches its iteration limit (default 10). Handle this by continuing the conversation:
```python Python
def handle_server_tool_conversation(client, user_query, tools, max_continuations=5):
"""
Handle server tool conversations that may require multiple continuations.
The server runs a sampling loop when executing server tools. If the loop
reaches its iteration limit, the API returns pause_turn. Continue the
conversation by sending the response back to let Claude finish.
"""
messages = [{"role": "user", "content": user_query}]
for _ in range(max_continuations):
response = client.messages.create(
model="claude-opus-5-5",
max_tokens=4096,
messages=messages,
tools=tools,
)
if response.stop_reason != "pause_turn":
# Claude finished processing - return the final response
return response
# pause_turn: replace the full message list to maintain alternating roles
messages = [
{"role": "user", "content": user_query},
{"role": "assistant", "content": response.content},
]
# Reached max continuations - return the last response
return response
```
```typescript TypeScript
async function handleServerToolConversation(
client: Anthropic,
userQuery: string,
tools: Anthropic.ToolUnion[],
maxContinuations = 5
): Promise {
let messages: Anthropic.MessageParam[] = [{ role: "user", content: userQuery }];
let response: Anthropic.Message;
for (let i = 0; i < maxContinuations; i++) {
response = await client.messages.create({
model: "claude-opus-5-5",
max_tokens: 4096,
messages,
tools
});
if (response.stop_reason !== "pause_turn") {
// Claude finished processing - return the final response
return response;
}
// pause_turn: replace the full message list to maintain alternating roles
messages = [
{ role: "user", content: userQuery },
{ role: "assistant", content: response.content }
];
}
// Reached max continuations - return the last response
return response!;
}
```
```csharp C#
static async Task HandleServerToolConversation(
AnthropicClient client,
string userQuery,
List tools,
int maxContinuations = 5)
{
List messages = [new() { Role = Role.User, Content = userQuery }];
Message response = null!;
for (var i = 0; i < maxContinuations; i++)
{
response = await client.Messages.Create(new MessageCreateParams
{
Model = Model.ClaudeOpus5_5,
MaxTokens = 4096,
Messages = messages,
Tools = tools
});
if (response.StopReason != "pause_turn")
{
// Claude finished processing - return the final response
return response;
}
// pause_turn: replace the full message list to maintain alternating roles
messages =
[
new() { Role = Role.User, Content = userQuery },
new()
{
Role = Role.Assistant,
Content = response.Content.Select(block => new ContentBlockParam(block.Json)).ToList()
}
];
}
// Reached max continuations - return the last response
return response;
}
```
```go Go
func handleServerToolConversation(
client anthropic.Client,
userQuery string,
tools []anthropic.ToolUnionParam,
maxContinuations int,
) (*anthropic.Message, error) {
messages := []anthropic.MessageParam{anthropic.NewUserMessage(anthropic.NewTextBlock(userQuery))}
var response *anthropic.Message
var err error
for range maxContinuations {
response, err = client.Messages.New(context.TODO(), anthropic.MessageNewParams{
Model: anthropic.ModelClaudeOpus5_5,
MaxTokens: 4096,
Messages: messages,
Tools: tools,
})
if err != nil {
return nil, err
}
if response.StopReason != "pause_turn" {
// Claude finished processing - return the final response
return response, nil
}
// pause_turn: replace the full message list to maintain alternating roles
var contentParams []anthropic.ContentBlockParamUnion
for _, block := range response.Content {
contentParams = append(contentParams, block.ToParam())
}
messages = []anthropic.MessageParam{
anthropic.NewUserMessage(anthropic.NewTextBlock(userQuery)),
anthropic.NewAssistantMessage(contentParams...),
}
}
// Reached max continuations - return the last response
return response, nil
}
```
```java Java
static Message handleServerToolConversation(
AnthropicClient client,
String userQuery,
List tools,
int maxContinuations
) {
Message response = null;
for (int i = 0; i < maxContinuations; i++) {
// Rebuild the params each iteration so messages aren't accumulated
MessageCreateParams.Builder params = MessageCreateParams.builder()
.model(Model.CLAUDE_OPUS_5_5)
.maxTokens(4096L)
.addUserMessage(userQuery);
tools.forEach(params::addTool);
if (response != null) {
params.addMessage(response);
}
response = client.messages().create(params.build());
if (!response.stopReason().map(StopReason.PAUSE_TURN::equals).orElse(false)) {
// Claude finished processing - return the final response
return response;
}
// pause_turn: loop again and send the response back
}
// Reached max continuations - return the last response
return response;
}
```
```php PHP
function handle_server_tool_conversation(
Client $client,
string $userQuery,
array $tools,
int $maxContinuations = 5
) {
$messages = [['role' => 'user', 'content' => $userQuery]];
$response = null;
for ($i = 0; $i < $maxContinuations; $i++) {
$response = $client->messages->create(
maxTokens: 4096,
messages: $messages,
model: 'claude-opus-5-5',
tools: $tools,
);
if ($response->stopReason !== 'pause_turn') {
// Claude finished processing - return the final response
return $response;
}
// pause_turn: replace the full message list to maintain alternating roles
$messages = [
['role' => 'user', 'content' => $userQuery],
['role' => 'assistant', 'content' => $response->content],
];
}
// Reached max continuations - return the last response
return $response;
}
```
```ruby Ruby
def handle_server_tool_conversation(client, user_query, tools, max_continuations: 5)
messages = [{ role: "user", content: user_query }]
response = nil
max_continuations.times do
response = client.messages.create(
model: "claude-opus-5-5",
max_tokens: 4096,
messages: messages,
tools: tools
)
# Claude finished processing - return the final response
return response unless response.stop_reason == :pause_turn
# pause_turn: replace the full message list to maintain alternating roles
messages = [
{ role: "user", content: user_query },
{ role: "assistant", content: response.content }
]
end
# Reached max continuations - return the last response
response
end
```
## Stop reasons vs. errors
It's important to distinguish between `stop_reason` values and actual errors:
### Stop reasons (successful responses)
* Part of the response body
* Indicate why generation stopped normally
* Response contains valid content
### Errors (failed requests)
* HTTP status codes 4xx or 5xx
* Indicate request processing failures
* Response contains error details
```bash cURL
# cURL exits non-zero on HTTP errors with --fail-with-body; inspect
# $? for errors and stop_reason for successful responses.
curl --fail-with-body -sS https://api.anthropic.com/v1/messages \
-H "x-api-key: $ANTHROPIC_API_KEY" \
-H "anthropic-version: 2023-06-01" \
-H "content-type: application/json" \
-d '{
"model": "claude-opus-5-5",
"max_tokens": 1024,
"messages": [{"role": "user", "content": "Hello!"}]
}' | jq '.stop_reason'
```
```bash CLI
# The CLI exits non-zero on API errors; stop_reason appears on success.
ant messages create \
--model claude-opus-5-5 \
--max-tokens 1024 \
--message '{role: user, content: "Hello!"}' \
--format json | jq '.stop_reason'
```
```python Python
client = anthropic.Anthropic()
try:
response = client.messages.create(
model="claude-opus-5-5",
max_tokens=1024,
messages=[{"role": "user", "content": "Hello!"}],
)
# Handle successful response with stop_reason
if response.stop_reason == "max_tokens":
print("Response was truncated")
except anthropic.APIStatusError as e:
# Handle actual errors
match e.status_code:
case 429:
print("Rate limit exceeded")
case 500:
print("Server error")
```
```typescript TypeScript
const client = new Anthropic();
try {
const response = await client.messages.create({
model: "claude-opus-5-5",
max_tokens: 1024,
messages: [{ role: "user", content: "Hello!" }]
});
// Handle successful response with stop_reason
if (response.stop_reason === "max_tokens") {
console.log("Response was truncated");
}
} catch (err) {
// Handle actual errors
if (err instanceof Anthropic.APIError) {
switch (err.status) {
case 429:
console.log("Rate limit exceeded");
break;
case 500:
console.log("Server error");
break;
}
} else {
throw err;
}
}
```
```csharp C#
AnthropicClient client = new();
try
{
var response = await client.Messages.Create(new MessageCreateParams
{
Model = Model.ClaudeOpus5_5,
MaxTokens = 1024,
Messages = [new() { Role = Role.User, Content = "Hello!" }]
});
// Handle successful response with stop_reason
if (response.StopReason == "max_tokens")
{
Console.WriteLine("Response was truncated");
}
}
catch (AnthropicRateLimitException)
{
// Handle actual errors
Console.WriteLine("Rate limit exceeded");
}
catch (Anthropic5xxException)
{
Console.WriteLine("Server error");
}
```
```go Go
client := anthropic.NewClient()
response, err := client.Messages.New(context.TODO(), anthropic.MessageNewParams{
Model: anthropic.ModelClaudeOpus5_5,
MaxTokens: 1024,
Messages: []anthropic.MessageParam{
anthropic.NewUserMessage(anthropic.NewTextBlock("Hello!")),
},
})
if err != nil {
// Handle actual errors
var apiErr *anthropic.Error
if errors.As(err, &apiErr) {
switch apiErr.StatusCode {
case 429:
fmt.Println("Rate limit exceeded")
case 500:
fmt.Println("Server error")
}
}
log.Fatal(err)
}
// Handle successful response with stop_reason
if response.StopReason == "max_tokens" {
fmt.Println("Response was truncated")
}
```
```java Java
AnthropicClient client = AnthropicOkHttpClient.fromEnv();
try {
Message response = client.messages().create(
MessageCreateParams.builder()
.model(Model.CLAUDE_OPUS_5_5)
.maxTokens(1024L)
.addUserMessage("Hello!")
.build()
);
// Handle successful response with stop_reason
if (response.stopReason().map(StopReason.MAX_TOKENS::equals).orElse(false)) {
IO.println("Response was truncated");
}
} catch (RateLimitException e) {
// Handle actual errors
IO.println("Rate limit exceeded");
} catch (AnthropicServiceException e) {
if (e.statusCode() == 500) {
IO.println("Server error");
}
}
```
```php PHP
$client = new Client();
try {
$response = $client->messages->create(
maxTokens: 1024,
messages: [['role' => 'user', 'content' => 'Hello!']],
model: 'claude-opus-5-5',
);
// Handle successful response with stop_reason
if ($response->stopReason === 'max_tokens') {
echo 'Response was truncated', PHP_EOL;
}
} catch (RateLimitException $e) {
// Handle actual errors
echo 'Rate limit exceeded', PHP_EOL;
} catch (InternalServerException $e) {
echo 'Server error', PHP_EOL;
}
```
```ruby Ruby
client = Anthropic::Client.new
begin
response = client.messages.create(
model: "claude-opus-5-5",
max_tokens: 1024,
messages: [{ role: "user", content: "Hello!" }]
)
# Handle successful response with stop_reason
if response.stop_reason == :max_tokens
puts "Response was truncated"
end
rescue Anthropic::Errors::RateLimitError
# Handle actual errors
puts "Rate limit exceeded"
rescue Anthropic::Errors::APIStatusError => e
puts "Server error" if e.status == 500
end
```
## Streaming considerations
When using streaming, `stop_reason` is:
* `null` in the initial `message_start` event
* Provided in the `message_delta` event
* Not provided in any other events
```bash cURL
# The message_delta event in the SSE stream carries stop_reason.
curl --no-buffer https://api.anthropic.com/v1/messages \
-H "x-api-key: $ANTHROPIC_API_KEY" \
-H "anthropic-version: 2023-06-01" \
-H "content-type: application/json" \
-d '{
"model": "claude-opus-5-5",
"max_tokens": 1024,
"stream": true,
"messages": [{"role": "user", "content": "Hello!"}]
}'
```
```bash CLI
# stop_reason appears in the message_delta event.
ant messages create --stream --format jsonl \
--model claude-opus-5-5 \
--max-tokens 1024 \
--message '{role: user, content: "Hello!"}' |
jq -c 'select(.type == "message_delta") | .delta.stop_reason'
```
```python Python
client = anthropic.Anthropic()
with client.messages.stream(
model="claude-opus-5-5",
max_tokens=1024,
messages=[{"role": "user", "content": "Hello!"}],
) as stream:
for event in stream:
if event.type == "message_delta":
stop_reason = event.delta.stop_reason
if stop_reason:
print(f"Stream ended with: {stop_reason}")
```
```typescript TypeScript
const client = new Anthropic();
const stream = client.messages.stream({
model: "claude-opus-5-5",
max_tokens: 1024,
messages: [{ role: "user", content: "Hello!" }]
});
for await (const event of stream) {
if (event.type === "message_delta" && event.delta.stop_reason) {
console.log(`Stream ended with: ${event.delta.stop_reason}`);
}
}
```
```csharp C#
AnthropicClient client = new();
var parameters = new MessageCreateParams
{
Model = Model.ClaudeOpus5_5,
MaxTokens = 1024,
Messages = [new() { Role = Role.User, Content = "Hello!" }]
};
await foreach (var streamEvent in client.Messages.CreateStreaming(parameters))
{
switch (streamEvent.Value)
{
case RawMessageDeltaEvent deltaEvent when deltaEvent.Delta.StopReason is not null:
Console.WriteLine($"Stream ended with: {deltaEvent.Delta.StopReason}");
break;
}
}
```
```go Go
client := anthropic.NewClient()
stream := client.Messages.NewStreaming(context.TODO(), anthropic.MessageNewParams{
Model: anthropic.ModelClaudeOpus5_5,
MaxTokens: 1024,
Messages: []anthropic.MessageParam{
anthropic.NewUserMessage(anthropic.NewTextBlock("Hello!")),
},
})
// Accumulate events into the final Message, which carries stop_reason.
message := anthropic.Message{}
for stream.Next() {
if err := message.Accumulate(stream.Current()); err != nil {
log.Fatal(err)
}
}
if err := stream.Err(); err != nil {
log.Fatal(err)
}
if message.StopReason != "" {
fmt.Printf("Stream ended with: %s\n", message.StopReason)
}
```
```java Java
AnthropicClient client = AnthropicOkHttpClient.fromEnv();
MessageCreateParams params = MessageCreateParams.builder()
.model(Model.CLAUDE_OPUS_5_5)
.maxTokens(1024L)
.addUserMessage("Hello!")
.build();
// Accumulate events into the final Message, which carries stop_reason.
MessageAccumulator accumulator = MessageAccumulator.create();
try (StreamResponse streamResponse =
client.messages().createStreaming(params)) {
streamResponse.stream().forEach(accumulator::accumulate);
}
accumulator.message().stopReason().ifPresent(stopReason ->
IO.println("Stream ended with: " + stopReason)
);
```
```php PHP
$client = new Client();
$stream = $client->messages->createStream(
maxTokens: 1024,
messages: [['role' => 'user', 'content' => 'Hello!']],
model: 'claude-opus-5-5',
);
foreach ($stream as $event) {
if ($event instanceof \Anthropic\Messages\RawMessageDeltaEvent && $event->delta->stopReason !== null) {
echo "Stream ended with: {$event->delta->stopReason}", PHP_EOL;
}
}
```
```ruby Ruby
client = Anthropic::Client.new
stream = client.messages.stream(
model: "claude-opus-5-5",
max_tokens: 1024,
messages: [{ role: "user", content: "Hello!" }]
)
stream.each do |event|
next unless event.type == :message_delta
stop_reason = event.delta.stop_reason
puts "Stream ended with: #{stop_reason}" if stop_reason
end
```
## Common patterns
### Handling tool use workflows
**Simpler with tool runner:** The following example shows manual tool handling. For most use cases, the [tool runner](https://platform.claude.com/docs/en/agents-and-tools/tool-use/tool-runner) automatically handles tool execution with much less code.
```python Python
def complete_tool_workflow(client, user_query, tools):
messages = [{"role": "user", "content": user_query}]
while True:
response = client.messages.create(
model="claude-opus-5-5",
max_tokens=1024,
messages=messages,
tools=tools,
)
if response.stop_reason == "tool_use":
# Execute tools and continue
tool_results = execute_tools(response.content)
messages.append({"role": "assistant", "content": response.content})
messages.append({"role": "user", "content": tool_results})
else:
# Final response
return response
```
```typescript TypeScript
async function completeToolWorkflow(
client: Anthropic,
userQuery: string,
tools: Anthropic.ToolUnion[]
): Promise {
const messages: Anthropic.MessageParam[] = [{ role: "user", content: userQuery }];
while (true) {
const response = await client.messages.create({
model: "claude-opus-5-5",
max_tokens: 1024,
messages,
tools
});
if (response.stop_reason === "tool_use") {
// Execute tools and continue
const toolResults = executeTools(response.content);
messages.push({ role: "assistant", content: response.content });
messages.push({ role: "user", content: toolResults });
} else {
// Final response
return response;
}
}
}
```
```csharp C#
static async Task CompleteToolWorkflow(
AnthropicClient client,
string userQuery,
List tools)
{
List messages = [new() { Role = Role.User, Content = userQuery }];
while (true)
{
var response = await client.Messages.Create(new MessageCreateParams
{
Model = Model.ClaudeOpus5_5,
MaxTokens = 1024,
Messages = messages,
Tools = tools
});
if (response.StopReason == "tool_use")
{
// Execute tools and continue
var toolResults = ExecuteTools(response.Content);
messages.Add(new()
{
Role = Role.Assistant,
Content = response.Content.Select(block => new ContentBlockParam(block.Json)).ToList()
});
messages.Add(new() { Role = Role.User, Content = toolResults });
}
else
{
// Final response
return response;
}
}
}
```
```go Go
func completeToolWorkflow(
client anthropic.Client,
userQuery string,
tools []anthropic.ToolUnionParam,
) (*anthropic.Message, error) {
messages := []anthropic.MessageParam{anthropic.NewUserMessage(anthropic.NewTextBlock(userQuery))}
for {
response, err := client.Messages.New(context.TODO(), anthropic.MessageNewParams{
Model: anthropic.ModelClaudeOpus5_5,
MaxTokens: 1024,
Messages: messages,
Tools: tools,
})
if err != nil {
return nil, err
}
if response.StopReason != "tool_use" {
// Final response
return response, nil
}
// Execute tools and continue
toolResults := executeTools(response.Content)
var contentParams []anthropic.ContentBlockParamUnion
for _, block := range response.Content {
contentParams = append(contentParams, block.ToParam())
}
messages = append(messages, anthropic.NewAssistantMessage(contentParams...))
messages = append(messages, anthropic.NewUserMessage(toolResults...))
}
}
```
```java Java
static Message completeToolWorkflow(
AnthropicClient client,
String userQuery,
List tools
) {
List messages = new ArrayList<>();
messages.add(MessageParam.builder().role(MessageParam.Role.USER).content(userQuery).build());
while (true) {
MessageCreateParams.Builder params = MessageCreateParams.builder()
.model(Model.CLAUDE_OPUS_5_5)
.maxTokens(1024L)
.messages(messages);
tools.forEach(params::addTool);
Message response = client.messages().create(params.build());
if (!response.stopReason().map(StopReason.TOOL_USE::equals).orElse(false)) {
// Final response
return response;
}
// Execute tools and continue
List toolResults = executeTools(response.content());
messages.add(response.toParam());
messages.add(MessageParam.builder()
.role(MessageParam.Role.USER)
.contentOfBlockParams(toolResults.stream().map(ContentBlockParam::ofToolResult).toList())
.build());
}
}
```
```php PHP
function complete_tool_workflow(Client $client, string $userQuery, array $tools)
{
$messages = [['role' => 'user', 'content' => $userQuery]];
while (true) {
$response = $client->messages->create(
maxTokens: 1024,
messages: $messages,
model: 'claude-opus-5-5',
tools: $tools,
);
if ($response->stopReason !== 'tool_use') {
// Final response
return $response;
}
// Execute tools and continue
$toolResults = execute_tools($response->content);
$messages[] = ['role' => 'assistant', 'content' => $response->content];
$messages[] = ['role' => 'user', 'content' => $toolResults];
}
}
```
```ruby Ruby
def complete_tool_workflow(client, user_query, tools)
messages = [{ role: "user", content: user_query }]
loop do
response = client.messages.create(
model: "claude-opus-5-5",
max_tokens: 1024,
messages: messages,
tools: tools
)
# Final response
return response unless response.stop_reason == :tool_use
# Execute tools and continue
tool_results = execute_tools(response.content)
messages << { role: "assistant", content: response.content }
messages << { role: "user", content: tool_results }
end
end
```
### Ensuring complete responses
```python Python
def get_complete_response(client, prompt, max_attempts=3):
messages = [{"role": "user", "content": prompt}]
full_response = ""
for _ in range(max_attempts):
response = client.messages.create(
model="claude-opus-5-5", messages=messages, max_tokens=4096
)
full_response += next(
(block.text for block in response.content if block.type == "text"), ""
)
if response.stop_reason != "max_tokens":
break
# Continue from where it left off
messages = [
{"role": "user", "content": prompt},
{"role": "assistant", "content": full_response},
{"role": "user", "content": "Please continue from where you left off."},
]
return full_response
```
```typescript TypeScript
async function getCompleteResponse(
client: Anthropic,
prompt: string,
maxAttempts = 3
): Promise {
let messages: Anthropic.MessageParam[] = [{ role: "user", content: prompt }];
let fullResponse = "";
for (let i = 0; i < maxAttempts; i++) {
const response = await client.messages.create({
model: "claude-opus-5-5",
max_tokens: 4096,
messages
});
const textBlock = response.content.find(
(block): block is Anthropic.TextBlock => block.type === "text"
);
fullResponse += textBlock?.text ?? "";
if (response.stop_reason !== "max_tokens") {
break;
}
// Continue from where it left off
messages = [
{ role: "user", content: prompt },
{ role: "assistant", content: fullResponse },
{ role: "user", content: "Please continue from where you left off." }
];
}
return fullResponse;
}
```
```csharp C#
static async Task GetCompleteResponse(AnthropicClient client, string prompt, int maxAttempts = 3)
{
List messages = [new() { Role = Role.User, Content = prompt }];
var fullResponse = "";
for (var i = 0; i < maxAttempts; i++)
{
var response = await client.Messages.Create(new MessageCreateParams
{
Model = Model.ClaudeOpus5_5,
MaxTokens = 4096,
Messages = messages
});
foreach (var block in response.Content)
{
if (block.TryPickText(out var textBlock))
{
fullResponse += textBlock.Text;
break;
}
}
if (response.StopReason != "max_tokens")
{
break;
}
// Continue from where it left off
messages =
[
new() { Role = Role.User, Content = prompt },
new() { Role = Role.Assistant, Content = fullResponse },
new() { Role = Role.User, Content = "Please continue from where you left off." }
];
}
return fullResponse;
}
```
```go Go
func getCompleteResponse(client anthropic.Client, prompt string, maxAttempts int) (string, error) {
messages := []anthropic.MessageParam{anthropic.NewUserMessage(anthropic.NewTextBlock(prompt))}
fullResponse := ""
for range maxAttempts {
response, err := client.Messages.New(context.TODO(), anthropic.MessageNewParams{
Model: anthropic.ModelClaudeOpus5_5,
MaxTokens: 4096,
Messages: messages,
})
if err != nil {
return "", err
}
for _, block := range response.Content {
if textBlock, ok := block.AsAny().(anthropic.TextBlock); ok {
fullResponse += textBlock.Text
break
}
}
if response.StopReason != "max_tokens" {
break
}
// Continue from where it left off
messages = []anthropic.MessageParam{
anthropic.NewUserMessage(anthropic.NewTextBlock(prompt)),
anthropic.NewAssistantMessage(anthropic.NewTextBlock(fullResponse)),
anthropic.NewUserMessage(anthropic.NewTextBlock("Please continue from where you left off.")),
}
}
return fullResponse, nil
}
```
```java Java
static String getCompleteResponse(AnthropicClient client, String prompt, int maxAttempts) {
List messages = List.of(
MessageParam.builder().role(MessageParam.Role.USER).content(prompt).build()
);
StringBuilder fullResponse = new StringBuilder();
for (int i = 0; i < maxAttempts; i++) {
Message response = client.messages().create(
MessageCreateParams.builder()
.model(Model.CLAUDE_OPUS_5_5)
.maxTokens(4096L)
.messages(messages)
.build()
);
response.content().stream()
.filter(ContentBlock::isText)
.findFirst()
.ifPresent(block -> fullResponse.append(block.asText().text()));
if (!response.stopReason().map(StopReason.MAX_TOKENS::equals).orElse(false)) {
break;
}
// Continue from where it left off
messages = List.of(
MessageParam.builder().role(MessageParam.Role.USER).content(prompt).build(),
MessageParam.builder().role(MessageParam.Role.ASSISTANT).content(fullResponse.toString()).build(),
MessageParam.builder().role(MessageParam.Role.USER).content("Please continue from where you left off.").build()
);
}
return fullResponse.toString();
}
```
```php PHP
function get_complete_response(Client $client, string $prompt, int $maxAttempts = 3): string
{
$messages = [['role' => 'user', 'content' => $prompt]];
$fullResponse = '';
for ($i = 0; $i < $maxAttempts; $i++) {
$response = $client->messages->create(
maxTokens: 4096,
messages: $messages,
model: 'claude-opus-5-5',
);
$fullResponse .= array_find($response->content, static fn ($block): bool => $block->type === 'text')?->text ?? '';
if ($response->stopReason !== 'max_tokens') {
break;
}
// Continue from where it left off
$messages = [
['role' => 'user', 'content' => $prompt],
['role' => 'assistant', 'content' => $fullResponse],
['role' => 'user', 'content' => 'Please continue from where you left off.'],
];
}
return $fullResponse;
}
```
```ruby Ruby
def get_complete_response(client, prompt, max_attempts: 3)
messages = [{ role: "user", content: prompt }]
full_response = +""
max_attempts.times do
response = client.messages.create(
model: "claude-opus-5-5",
max_tokens: 4096,
messages: messages
)
full_response << response.content.find { it.type == :text }&.text.to_s
break unless response.stop_reason == :max_tokens
# Continue from where it left off
messages = [
{ role: "user", content: prompt },
{ role: "assistant", content: full_response },
{ role: "user", content: "Please continue from where you left off." }
]
end
full_response
end
```
### Getting maximum tokens without knowing input size
With the `model_context_window_exceeded` stop reason, you can request the maximum possible tokens without calculating input size:
```python Python
def get_max_possible_tokens(client, prompt):
"""
Get as many tokens as possible within the model's context window
without needing to calculate input token count
"""
response = client.beta.messages.create(
model="claude-opus-5-5",
messages=[{"role": "user", "content": prompt}],
max_tokens=20000, # Python SDK requires streaming for max_tokens above ~21k
)
match response.stop_reason:
case "model_context_window_exceeded":
# Got the maximum possible tokens given input size
print(
f"Generated {response.usage.output_tokens} tokens (context limit reached)"
)
case "max_tokens":
# Got exactly the requested tokens
print(
f"Generated {response.usage.output_tokens} tokens (max_tokens reached)"
)
case _:
# Natural completion
print(
f"Generated {response.usage.output_tokens} tokens (natural completion)"
)
return next((block.text for block in response.content if block.type == "text"), "")
```
```typescript TypeScript
async function getMaxPossibleTokens(client: Anthropic, prompt: string): Promise {
const response = await client.beta.messages.create({
model: "claude-opus-5-5",
max_tokens: 20000,
messages: [{ role: "user", content: prompt }]
});
const tokens = response.usage.output_tokens;
switch (response.stop_reason) {
case "model_context_window_exceeded":
// Got the maximum possible tokens given input size
console.log(`Generated ${tokens} tokens (context limit reached)`);
break;
case "max_tokens":
// Got exactly the requested tokens
console.log(`Generated ${tokens} tokens (max_tokens reached)`);
break;
default:
// Natural completion
console.log(`Generated ${tokens} tokens (natural completion)`);
}
const textBlock = response.content.find(
(block): block is Anthropic.Beta.BetaTextBlock => block.type === "text"
);
return textBlock?.text ?? "";
}
```
```csharp C#
using Anthropic.Models.Beta.Messages;
using Model = Anthropic.Models.Messages.Model;
static async Task GetMaxPossibleTokens(AnthropicClient client, string prompt)
{
var response = await client.Beta.Messages.Create(new MessageCreateParams
{
Model = Model.ClaudeOpus5_5,
MaxTokens = 20000,
Messages = [new() { Role = Role.User, Content = prompt }]
});
var tokens = response.Usage.OutputTokens;
var reason = response.StopReason?.Value();
if (reason == BetaStopReason.ModelContextWindowExceeded)
{
// Got the maximum possible tokens given input size
Console.WriteLine($"Generated {tokens} tokens (context limit reached)");
}
else if (reason == BetaStopReason.MaxTokens)
{
// Got exactly the requested tokens
Console.WriteLine($"Generated {tokens} tokens (max_tokens reached)");
}
else
{
// Natural completion
Console.WriteLine($"Generated {tokens} tokens (natural completion)");
}
return response.Content.Select(b => b.Value).OfType().FirstOrDefault()?.Text ?? "";
}
```
```go Go
func getMaxPossibleTokens(client anthropic.Client, prompt string) (string, error) {
response, err := client.Beta.Messages.New(context.TODO(), anthropic.BetaMessageNewParams{
Model: anthropic.ModelClaudeOpus5_5,
MaxTokens: 20000,
Messages: []anthropic.BetaMessageParam{
anthropic.NewBetaUserMessage(anthropic.NewBetaTextBlock(prompt)),
},
})
if err != nil {
return "", err
}
tokens := response.Usage.OutputTokens
switch response.StopReason {
case anthropic.BetaStopReasonModelContextWindowExceeded:
// Got the maximum possible tokens given input size
fmt.Printf("Generated %d tokens (context limit reached)\n", tokens)
case anthropic.BetaStopReasonMaxTokens:
// Got exactly the requested tokens
fmt.Printf("Generated %d tokens (max_tokens reached)\n", tokens)
default:
// Natural completion
fmt.Printf("Generated %d tokens (natural completion)\n", tokens)
}
for _, block := range response.Content {
if textBlock, ok := block.AsAny().(anthropic.BetaTextBlock); ok {
return textBlock.Text, nil
}
}
return "", nil
}
```
```java Java
import com.anthropic.models.beta.messages.BetaContentBlock;
import com.anthropic.models.beta.messages.BetaMessage;
import com.anthropic.models.beta.messages.BetaStopReason;
import com.anthropic.models.beta.messages.MessageCreateParams;
static String getMaxPossibleTokens(AnthropicClient client, String prompt) {
BetaMessage response = client.beta().messages().create(
MessageCreateParams.builder()
.model(Model.CLAUDE_OPUS_5_5)
.maxTokens(20000L)
.addUserMessage(prompt)
.build()
);
long tokens = response.usage().outputTokens();
BetaStopReason reason = response.stopReason().orElse(BetaStopReason.END_TURN);
if (reason.equals(BetaStopReason.MODEL_CONTEXT_WINDOW_EXCEEDED)) {
// Got the maximum possible tokens given input size
IO.println("Generated " + tokens + " tokens (context limit reached)");
} else if (reason.equals(BetaStopReason.MAX_TOKENS)) {
// Got exactly the requested tokens
IO.println("Generated " + tokens + " tokens (max_tokens reached)");
} else {
// Natural completion
IO.println("Generated " + tokens + " tokens (natural completion)");
}
return response.content().stream()
.filter(BetaContentBlock::isText)
.findFirst()
.map(block -> block.asText().text())
.orElse("");
}
```
```php PHP
function get_max_possible_tokens(Client $client, string $prompt): string
{
$response = $client->beta->messages->create(
maxTokens: 20000,
messages: [['role' => 'user', 'content' => $prompt]],
model: 'claude-opus-5-5',
);
$tokens = $response->usage->outputTokens;
echo match ($response->stopReason) {
// Got the maximum possible tokens given input size
'model_context_window_exceeded' => "Generated {$tokens} tokens (context limit reached)",
// Got exactly the requested tokens
'max_tokens' => "Generated {$tokens} tokens (max_tokens reached)",
// Natural completion
default => "Generated {$tokens} tokens (natural completion)",
}, PHP_EOL;
return array_find($response->content, static fn ($block): bool => $block->type === 'text')?->text ?? '';
}
```
```ruby Ruby
def get_max_possible_tokens(client, prompt)
response = client.beta.messages.create(
model: "claude-opus-5-5",
max_tokens: 20000,
messages: [{ role: "user", content: prompt }]
)
tokens = response.usage.output_tokens
case response.stop_reason
when :model_context_window_exceeded
# Got the maximum possible tokens given input size
puts "Generated #{tokens} tokens (context limit reached)"
when :max_tokens
# Got exactly the requested tokens
puts "Generated #{tokens} tokens (max_tokens reached)"
else
# Natural completion
puts "Generated #{tokens} tokens (natural completion)"
end
response.content.find { it.type == :text }.text
end
```
## Next steps
Retry refused requests on a fallback model, server-side or in your client.
Let the SDK manage the `tool_use` loop, result formatting, and retries for you.
Read `stop_reason` from the `message_delta` event when streaming.
Handle 4xx and 5xx HTTP errors, which are distinct from stop reasons.