Skip to main content
Reflection models can reason before they answer: they generate reasoning tokens to work through a problem, then write the final answer. More reasoning tends to improve answers to hard problems, such as math, code, and multi-step planning, at the cost of latency and tokens.

Set reasoning effort

Beam-1.0-501B-A23B always reasons. Use reasoning_effort to choose how much. It supports these values:
Any other value returns a 400 error with code unsupported_value, and the message lists the values the model accepts. This includes OpenAI effort names that the model doesn’t support, such as none and minimal. An integer effort returns a 400 error with code invalid_type. Lower the effort to low or medium for simple tasks where latency matters more, and raise it for hard problems where answer quality matters most.

Discover supported efforts

The models endpoints describe each model’s efforts in its reasoning object:
Model object
A model that doesn’t announce its efforts has no reasoning object.

Read the reasoning

The model’s reasoning is returned separately from its answer, in message.reasoning_content. The answer is in message.content as usual.
Response message
When streaming, reasoning arrives in delta.reasoning_content, before the answer arrives in delta.content.
Reasoning is useful for debugging and for showing progress, but it is not written for end users. Show content as the answer.

Reasoning tokens and limits

Reasoning tokens are counted in usage.completion_tokens and reported separately in usage.completion_tokens_details.reasoning_tokens. They count toward max_completion_tokens and toward rate limits. If max_completion_tokens is too low, the model can run out of tokens while reasoning. The response then has finish_reason: "length" and content may be null. Raise max_completion_tokens or lower reasoning_effort.

Reasoning in multi-turn conversations

By default, reasoning_content is required on every assistant message in a request, as a non-empty string: send each prior turn back exactly as you received it, with its reasoning. During tool calling, return the assistant message with its tool_calls and reasoning_content unchanged. An assistant message without it returns a 400 error with code missing_required_parameter, unless the request opts in to messages without reasoning. The model occasionally returns a tool-call turn without reasoning, which can’t be sent back by default; see Troubleshooting.
Assistant message in a request

Assistant messages without reasoning

If your client can’t send back every assistant message with its reasoning, set the Reflection-Missing-Reasoning-Strategy request header to auto. The API then accepts assistant messages without reasoning_content and chooses how to make up for the missing reasoning when needed, which can lower the quality of the response. Send the header with each request that needs it.
Last modified on November 18, 2026