Which model returns this 400?
claude-sonnet-5-5. Anthropic's migration guide says a non-default temperature, top_p, or top_k returns a 400 error on that model. Claude Sonnet 4.6 and earlier, and Claude Haiku 4.5, still accept those fields.
temperature is deprecated for this model
A request to claude-sonnet-5-5 that sets a non-default temperature, top_p, or top_k comes back as HTTP 400. Remove those fields.
API version 2026-10-21: sending temperature, top_p, or top_k returns a 400 on every model, even at the default value. API version 2026-10-21 sampling parameters. Source: Claude API versioning, checked 2026-10-22.
Error text
temperature is deprecated for this modelAnthropic's Sonnet 5.5 migration guide says a non-default temperature, top_p, or top_k returns a 400 error. Remove them. Claude Sonnet 4.6 and earlier, and Claude Haiku 4.5, still accept those fields. Source: Claude Sonnet 5.5 migration guide, checked October 20, 2026.
Affected models
| Model id | What the source says |
|---|---|
claude-sonnet-5-5 | Returns 400 when temperature, top_p, or top_k is set to a non-default value. |
Fix
Raw SDK
Call the Messages API without temperature, top_p, or top_k. The migration guide's working request leaves those fields out.
Python
import anthropic
client = anthropic.Anthropic()
message = client.messages.create(
model="claude-sonnet-5-5",
max_tokens=16000,
messages=[{"role": "user", "content": "Hello"}],
)
TypeScript
import Anthropic from "@anthropic-ai/sdk";
const client = new Anthropic();
const message = await client.messages.create({
model: "claude-sonnet-5-5",
max_tokens: 16000,
messages: [{ role: "user", content: "Hello" }],
});
Source: Claude Sonnet 5.5 migration guide, checked October 20, 2026.
LiteLLM
Do not send temperature, top_p, or top_k. LiteLLM's drop_params setting removes a field only when LiteLLM already lists that field as unsupported for the model. To remove temperature and top_p on this call, pass additional_drop_params.
Python
import litellm
response = litellm.completion(
model="anthropic/claude-sonnet-5-5",
messages=[{"role": "user", "content": "Hello"}],
additional_drop_params=["temperature", "top_p"],
)
Source: LiteLLM drop unsupported params, checked October 20, 2026.
Vercel AI SDK
Leave temperature, topP, and topK unset on generateText and streamText. The AI SDK passes those values through to the provider.
TypeScript
import { anthropic } from "@ai-sdk/anthropic";
import { generateText } from "ai";
const { text } = await generateText({
model: anthropic("claude-sonnet-5-5"),
prompt: "Hello",
});
Source: AI SDK generateText, checked October 20, 2026.
Switch guides
More request errors
- temperature, top_p, and top_k return 400 on API version 2026-10-21
- To turn thinking off on this model, send "thinking": {"type": "between_tools"} instead of {"type": "disabled"}. The model does not think before responding. The short updates it writes between tool calls come back as thinking blocks.
- "thinking.type.enabled" is not supported for this model. Use "thinking.type.adaptive" and "output_config.effort" to control thinking behavior.
- tool_choice: type "tool" and "any" are not supported for this model.
- This model does not support assistant message prefill. The conversation must end with a user message.
- top_p and top_k return a 400 on claude-sonnet-5-5
Questions
What should the request send instead?
Omit temperature, top_p, and top_k. The same guide says thinking budgets are replaced by output_config.effort. The switch guide lists each request change: claude-sonnet-4-5-20250929 to claude-sonnet-5-5.