Skip to content
DriftCaught
AnthropicNew version

claude-sonnet-4-5-20250929

New version, announced 29 September 2025. Rerun the suite: Add handling for the model_context_window_exceeded stop reason so long inputs fail visibly, and rerun the long context cases that previously truncated.

What Anthropic says changed

Anthropic launched claude-sonnet-4-5-20250929, which it describes as its best model for complex agents and coding. The notes state that a new stop reason model_context_window_exceeded was added, that the memory tool launched in beta and that context editing launched in beta.

Anthropic: Claude API release notes

announced 2025-09-29 . verified 2026-09-22

This section restates the provider. It adds no judgement the provider did not make, and it contains no measurement of ours.

Entry

Provider
Anthropic
Models
claude-sonnet-4-5-20250929
Change type
New version
Announced
2025-09-29
Verified
2026-09-22

What this kind of change typically touches

  • RAG

    Context editing and a new stop reason change what actually reaches the model on a long input. Ragas defines faithfulness as the share of claims in a response that can be inferred from the retrieved context, so the score follows whatever survived.

    Ragas

  • Agent

    A memory tool adds calls that the final answer never shows. Arize Phoenix reconstructs tool_path, the ordered tool calls with their arguments, and tool_io, what each tool was called with and what it returned.

    Arize AI

What to do this week

Rerun the suite

Add handling for the model_context_window_exceeded stop reason so long inputs fail visibly, and rerun the long context cases that previously truncated.

If you are on a plan

A tracked provider update triggers a rerun of your suite within 2 business days of the announcement, and you get a written verdict. That is one of the six guarantees on every plan.

See plansRead EV-RERUN

Other Anthropic changes

AnthropicNew version

claude-fable-5-1, claude-mythos-5-1

Anthropic launched claude-fable-5-1 and claude-mythos-5-1 with a 1M token context window by default, 128k maximum output tokens and always on adaptive thinking, priced at $10 and $50 per million tokens. The notes state that tool_choice types any and tool are not supported and return a 400 error, and that thinking blocks are preserved only for the same model or a newer one.

Revisit the promptSearch your codebase for tool_choice values of any or tool, since those requests now fail, and check any code that replays thinking blocks across models.

Source: Anthropic, Claude API release notes . verified 2026-09-22

AnthropicDeprecation

temperature, top_p, top_k

Anthropic released v1.0 of the Python SDK. It requires Python 3.10 or later and removes long deprecated surface, including the legacy Text Completions API and the temperature, top_p and top_k parameters on Messages methods. Anthropic's deprecations page states that passing them on v1.0 and later raises a TypeError.

Plan a migrationSearch your code for temperature, top_p and top_k on Messages calls and remove them, and pin your Python SDK version until that change is made.

Source: Anthropic, Claude API release notes . verified 2026-09-22

AnthropicNew version

claude-opus-5, claude-opus-4-7

Anthropic launched claude-opus-5 at $5 and $25 per million tokens, with a 1M token context window as both the default and the maximum and 128k maximum output tokens. The notes state that thinking is on by default, that disabling thinking is only allowed at effort high or below, which they call a breaking change from Claude Opus 4.8, and that Fast mode was removed for Claude Opus 4.7 so those requests return an error.

Revisit the promptFind any request that disables thinking at effort xhigh or max, since that combination is now rejected, and remove the fast mode flag from Claude Opus 4.7 calls.

Source: Anthropic, Claude API release notes . verified 2026-09-22

Questions

How do I know this is accurate?
The entry links to Anthropic's own page, and the date we read it is on this page. If the provider's page and this page disagree, the provider's page is right and we want to hear about it.
Does this mean my feature broke?
No. It means a thing your feature depends on changed. Whether anything about your output changed is a question only a rerun against your own dataset can answer, and that is what the recommended action is for.
Why is there no score or benchmark here?
Because we publish none. This site restates provider announcements and cites published practitioner guides. Any measurement we make belongs to the customer whose dataset produced it.

Sources

Change-log entries restate provider announcements with sources and dates. Impact notes cite public practitioner reports. The site publishes no measurements of its own on any model.

Related