Claude Fable 5.1 Release Details and Evaluation
Anthropic released Claude Fable 5.1 on September 1, 2026. This model features a 1M token context window, 128k max output, and specific API pricing.
Official Announcement and Specifications
Anthropic announced the release of Claude Fable 5.1 (claude-fable-5-1) on September 1, 2026. The source identifies this model as the successor to Claude Fable 5, intended for long-running agentic coding, knowledge work, and research. The announcement also notes the simultaneous release of Claude Mythos 5.1 for Project Glasswing participants.
The documented specifications for Claude Fable 5.1 include a default context window of 1 million tokens and a maximum output limit of 128,000 tokens. The model operates with always-on adaptive thinking. Input modalities are listed as text and image, while output modality is text. Open weights are not confirmed for this release.
Pricing Structure and Availability
The source states that Claude Fable 5.1 carries a price of $10 USD per million input tokens and $50 USD per million output tokens. These rates are identical to those of Claude Fable 5. A change in cost structure applies to cache reads, which are now priced at $0.25 per million tokens.
Availability is confirmed for the Claude API, Claude in Amazon Bedrock, Claude Platform on AWS, Claude on Google Cloud, and Claude in Microsoft Foundry. The source does not provide information regarding local installation requirements or alternative hosting arrangements beyond these listed platforms.
Evaluation Procedure for Engineers
Engineers can evaluate the model by designing tests that validate its performance against specific project requirements. Since the source confirms the context window size but not comparative speed or quality metrics, evaluation must rely on direct measurement.
A practical procedure involves testing the model with inputs approaching the 1 million token limit to measure latency and accuracy. Developers should compare outputs from Claude Fable 5.1 against their existing baselines using standardized datasets. Validating the effectiveness of the always-on adaptive thinking feature requires measuring task completion rates on complex agentic coding tasks. Finally, teams should calculate total costs based on the $10/$50 per MTok rates and the $0.25 per MTok cache read rate to determine budgetary impact.
- Test input processing with documents up to 1M tokens.
- Measure response time for 128k token outputs.
- Compare code generation accuracy against previous internal benchmarks.
- Validate cost efficiency using the updated cache read pricing.
Artificial Analysis benchmark review
Benchmark review updated on . These scores apply to the linked artifact and recorded methodology; they do not establish the performance of a different runtime or quantized model.
| Measurement | Recorded result | Measured |
|---|---|---|
| Artificial Analysis Coding Index (v4.1) ↗ | 81.6 index_points | 2026-09-02 |
| Artificial Analysis Coding Index (v4.2) ↗ | 81.6 index_points | 2026-09-05 |
| Artificial Analysis Coding Index (v4.3) ↗ | 81.6 index_points | 2026-09-09 |
| Artificial Analysis Intelligence Index (v4.1) ↗ | 65.7 index_points | 2026-09-02 |
| Artificial Analysis Intelligence Index (v4.2) ↗ | 56.8 index_points | 2026-09-05 |
| Artificial Analysis Intelligence Index (v4.3) ↗ | 53.4 index_points | 2026-09-09 |
Review the full benchmark provenance and source methodology →
Primary sources
Frequently asked questions
What is the exact date of the Claude Fable 5.1 announcement?
The official announcement occurred on September 1, 2026. This date distinguishes the historical event from the current publication date of this article.
Does the source confirm open weights availability?
No. The source records that open weights are not confirmed. It does not establish whether local installation is possible or if API access is mandatory.