Is Claude Code switching my model?

It can, and the transcripts on your machine record it: Claude Code writes the model that answered on every response, and a notice when it moves a session to another model. What you do not get from Claude Code is the history: how often it happened, on which days and client versions, and how many responses the other model served before the session came back. npx --allow-git=root github:agentwares/which-model reads that from the transcripts already in ~/.claude/projects, lists every switch in a session's main thread with what was recorded just before it, and sends nothing.

What the transcript records

The method

A switch is two consecutive responses in one session's main thread answered by different models. Subagents are left out, because they run on their own models by design. Each switch is labelled by what the transcript records between the two responses:

For each switch it gives the date, the two models, the client version, and how many responses and hours the new model then served. Once a requested model is recorded (your /model, or a notice's original model), it also counts the responses answered by a different one.

$ npx --allow-git=root github:agentwares/which-model
Served-model switches nothing you did explains (Claude Code, main thread): 2 in 1 session
  16 Jul 02:42  session 63c29f12  Claude Code 2.1.197
    claude-fable-5 → claude-opus-4-8 after a fallback notice the client recorded (claude-fable-5 →
    claude-opus-4-8, trigger: refusal); served 308 responses over 21.5 h, until the next switch.
  17 Jul 00:15  session 63c29f12  Claude Code 2.1.197
    claude-opus-4-8 → claude-fable-5 with nothing recorded before it (no /model, no fallback notice,
    same client version); served 12 responses over 0.8 h, to the end of the session.
  Not listed: 1 after your own /model.
  Answered by a different model than the one last recorded as requested: 308 responses in 1 session.

What Claude Code shows, as of 8 October 2026

The docs say a session can fall back to another model and that a notice names both models when it does. The changelog shows the notice still changing: 2.1.285 fixed /cost reporting a turn under the wrong model after a fallback, and 2.1.286 made the notice say when a fallback dropped the context window from 1M to 200K tokens. There is no documented label of the answering model on each response and nothing that counts a session's time on the other model; that is what #78888 asks for.

What people reported

In #78888 (18 July 2026) a user found from the per-message model field that one session was served by claude-opus-4-8 for about 840 messages over about 23 hours while the interface still showed Fable 5. #83510 (3 August 2026) reports under-disclosed fallback from Fable 5 to Opus 4.8 alongside measured quality changes. Both were found by hand, from the same fields which-model reads.

What it cannot tell you

It sees your machine, not Anthropic's servers, so it never says why a switch happened. A switch with nothing recorded before it may be a model you picked in a way the transcript does not record. And it cannot do the same for Codex: Codex records the model it requested, not the one that answered.

$ npx --allow-git=root github:agentwares/which-model