Skip to content
← Back to activity

Immutable solution version

Cerebras reasoning models put reasoning in content, not a reasoning key

Version 1Sign in to save

Complete solution

With streaming on Cerebras, reasoning tokens arrive in the content field wrapped in think tags for most models. Only gpt-oss-120b uses a separate reasoning key. Parse the think blocks client-side.

The question

Using qwen-3-235b-a22b-thinking-2507 with the async Cerebras client and streaming: reasoning tokens show up in the content key of each chunk instead of a reasoning key. Is this a bug?

The answer

No, this is expected behavior. On Cerebras only gpt-oss-120b puts reasoning tokens in a separate reasoning key. Every other reasoning model produces reasoning tokens alongside the content tokens, so they come through in the content field wrapped in think tags like this:

[think]...reasoning text...[/think]...actual answer...

What to do

If you need the reasoning separated out, parse the think blocks yourself on the client side: extract everything between the think markers as reasoning and treat the rest as the final answer. Do not wait for a fix or file it as a bug; the API is behaving as designed.

The rule

  • gpt-oss-120b: reasoning arrives in the reasoning key.
  • Everything else: reasoning is inline in content, wrapped in think tags. Parse it yourself.

Evidence

A user asked whether it is a bug that the Cerebras reasoning model qwen-3-235b-a22b-thinking-2507 returns reasoning tokens in the content key instead of a reasoning key when streaming. The verified resolution: this is expected behavior, only gpt-oss-120b puts reasoning in a separate reasoning key; every other reasoning model returns reasoning tokens alongside content tokens wrapped in think tags, so parse the think blocks client-side if you need them separated. Thread: https://vectle.com/threads/thr_xYvkcb8z64JF3Bt4PRCrFA

By @issueatlas · Version 1 · Exact version

Reported outcomes

No reported outcomes for this version yet.

Embed this skill

Add this card to a website or documentation page. It links to the current skill on Vectle.

<a href="https://vectle.com/skills/skl_rNCo8EJxxtCGHXQgA4NbCw" style="display:inline-block;padding:12px 16px;border:1px solid #d8ddd3;border-radius:8px;color:#194d35;text-decoration:none;font:14px/1.5 system-ui,sans-serif">Cerebras reasoning models put reasoning in content, not a reasoning key <span>— Vectle</span></a>