Day 0 Claude Fable 5.1 release

[misbahsy]

Day 0 support for Claude Fable 5.1

Call it through the LiteLLM AI Gateway using the same OpenAI-compatible request you already use, with spend tracking, rate limits, fallbacks, and logging in one place. Use it with Anthropic, Bedrock, Gemini Enterprise Agent Platform, and Azure AI.

What's new:

• Cache reads cost a quarter of what they did: $0.25 / MTok, down from $1.00 on Fable 5. That is 0.025x the base input price, where every other Claude model sits at 0.1x. Input and output are unchanged at $10 / MTok and $50 / MTok, cache writes stay at $12.50 (5-minute) and $20 (1-hour), and the 512-token cache minimum carries over. Anthropic puts the effect at roughly 25% lower cost on typical workloads and up to 45% on agentic ones.

• Stronger on long-running work: agentic coding across multi-hour sessions, documents and spreadsheets and slides, multistep research, dense PDFs and charts, and computer use. The gap over Fable 5 is widest at higher effort.

• Same shape as Fable 5 otherwise: 1M-token context window, 128K max output, adaptive thinking always on, effort ladder from low to max with high as the default.

Get started, no redeploy needed:

PR #39148 registers claude-fable-5-1, prices cache reads at $0.25 / MTok, and unblocks reasoning_effort, which returned a 400 before the model was in the registry. Reload your pricing in the UI under Models + Endpoints → Price Data → Reload Price Data (or POST /reload/model_cost_map as an admin) to pick it up without a redeploy.

model_list:
  - model_name: claude-fable-5-1
    litellm_params:
      model: anthropic/claude-fable-5-1
      api_key: os.environ/ANTHROPIC_API_KEY

Running with LITELLM_LOCAL_MODEL_COST_MAP=true? ghcr.io/berriai/litellm:v1.101.0-dev.1 will be released today, and that version (or any later version) adds Fable 5.1 support

Read the full guide → Day 0 Support: Claude Fable 5.1

Mateo
AI Engineer @ LiteLLM

Misbah
Dev Rel @ LiteLLM