Follow Discord
Sweep 08 Oct 2026 · 18:53Z Build v2.1.295 516 read Stable v2.1.286 Latest v2.1.295 Next v2.1.295 Feeds RSS JSON llms.txt llms-full.txt Unofficial
HomeReleasesv2.1.293entry #1

Claude Code v2.1.293 ·

Claude Haiku 5.5 added, and the haiku alias points to it on the first-party API

Claude Haiku 5.5 joins the model list and the haiku alias now resolves to it on Anthropic's API, while Bedrock, Vertex and other providers stay on Haiku 4.5

Group of 5 You'll notice Notable Improvements
JSON All of v2.1.293
You'll noticeTier: how much it should matter to you
5Useful: my rating, 1 to 5
4Signal: worth watching, 1 to 5
ModelsArea: what it touches
ImprovementsKind: in v2.1.293,
What probably matters to youSection of the release

What

This build knows about a new model, Claude Haiku 5.5 (claude-haiku-5-5). The name appears 16 times in this build and not at all in the previous one.

  • The haiku alias, the short name that stands for the current Haiku model, now defaults to claude-haiku-5-5 in place of claude-haiku-4-5. A per_provider map keeps bedrock, vertex, foundry, mantle, anthropic_aws, anthropic_google_cloud and gateway on claude-haiku-4-5, so only Anthropic's own API gets the new model.
  • The built-in model entry lists a 1M-token context window, 128000 maximum output tokens, default effort medium, a Bedrock id, and claude-haiku-4-5 as the fallback on third-party providers.
  • A haiku_55 pricing tier is added, listed as input 0.1 and output 0.5, with a long-prompt surcharge above 100,000 prompt tokens.
  • The short name haiku-5-5 maps to the model, any id containing claude-haiku-5-5 is recognised as it, and it gets its own family entry, haiku55.
  • A new environment variable, VERTEX_REGION_CLAUDE_HAIKU_5_5, sets which Google Vertex region this model uses.
  • A capability check that treated any model name containing "haiku" like claude-opus-4-6 and claude-sonnet-4-6 now leaves claude-haiku-5-5 out of that group.
  • The model-list text in the system prompt now says "The most recent Claude models are the Claude 5 family." without "and Haiku 4.5", and no longer maps claude-haiku-4-5 to its dated id.
  • New system-prompt guidance tells the model that reasoning effort does not change how much of a request it should finish, and that it should not check in before starting a large task. It applies only to models that pass the haiku_5_5_early_stopping_guidance check and when the server-side flag tengu_idempotent_wolf is on. The flag defaults to true in the code, and nothing has been read about it.

Why

If you use the haiku alias on Anthropic's own API, including for background and side tasks that use it, those requests now go to Haiku 5.5 and are priced as Haiku 5.5. On Bedrock, Vertex, Foundry, Mantle or a gateway nothing moves unless you choose the new model yourself.

Read from
Names in the bundlehaikuclaude-haiku-5-5
Since it was published

The entry above is what we published on the day. These lines were added later, as Anthropic's own pages caught up, and they sit beside the original rather than replacing it.

Confirmed since Anthropic's documentation has since written up haiku, on Model configuration. On the Anthropic API, the 1M context window uses standard model pricing with no premium for tokens beyond 200K, except on Haiku 5.5, which [costs more on prompts longer than 100K tokens](#haiku-5-5-context-window-and-pricing). For plans wh… model-config see the edit
How sure we are
Two sources agreeTwo things we can check say the same as this entry.
Anthropic's documentation agreesAnthropic's documentation has since written up haiku, on Model configuration.
Anthropic's release notes agreeAdded Claude Haiku 5.5 (claude-haiku-5-5), now the default Haiku model on the Anthropic API — 1M context, $0.10/$0.50 per Mtok ($0.50/$2.50…

See this entry in the whole of v2.1.293 →

Feedback