What is Claude Fable 5.1 best suited for?
It is designed for demanding reasoning, codebase-wide engineering, long-running unattended agents, multistep research, and complex office-document workflows.
CL
Claude Fable 5.1
Also known as: Mythos 5.1
Release date: 2026-09-01Updated: 2026-09-02Knowledge cutoff: 2026-06Views: 125
Parameters
No data
Context length
1M
Multilingual
Supported
Reasoning ability
No data
Claude Fable 5.1 is Anthropic's generally available flagship model, released on September 1, 2026 for demanding reasoning, long-horizon agentic coding, research, and complex knowledge work. It supports a 1M-token context window, 128K maximum output, text and image input, and always-on adaptive thinking. Claude API pricing is $10 per million input tokens, $50 per million output tokens, and $0.25 for cache reads. Parameter count and training compute are undisclosed.
Data sourced primarily from official releases (GitHub, Hugging Face, papers), then benchmark leaderboards, then third-party evaluators. Learn about our data methodology
Claude Fable 5.1
Model basics
Reasoning traces
Supported
Thinking modes
Thinking Level · Adaptive Thinking (Default)
Context length
1M tokens
Max output length
128K tokens
Model type
Reasoning model
Modality (in / out)
Text, Image → Text
Release date
2026-09-01
Model file size
No data
MoE architecture
No
Total params / Active params
No data / Not applicable
Knowledge cutoff
2026-06
Claude Fable 5.1
Open source & experience
Claude Fable 5.1
Official resources
DataLearnerAI blog
N/A
Claude Fable 5.1
API details
API speed
No data
💡Default unit: $/1M tokens. If vendors use other units, follow their published pricing.
Standard
| Type | Condition | Input | Output |
|---|---|---|---|
| Text | - | $10.00/ 1M | $50.00/ 1M |
Batch
| Type | Condition | Input | Output |
|---|---|---|---|
| Text | - | $5.00/ 1M | $25.00/ 1M |
Cache PricingPrompt Cache
| Type | TTL | Write | Read |
|---|---|---|---|
| Text | 5m | $12.50/ 1M | $0.250/ 1M |
| Text | 1h | $20.00/ 1M | — |
“—” means the modality is not billed in that direction, or the vendor has not published a price for it.
Claude Fable 5.1
Benchmark Results
Claude Fable 5.1 currently shows benchmark results led by HLE (1 / 188, score 65), ARC-AGI-1 (3 / 90, score 97.50), ARC-AGI-2 (3 / 84, score 90). This page also consolidates core specs, context limits, and API pricing so you can evaluate the model from benchmark results and deployment constraints together.
General Knowledge
17 evaluationsBenchmark / mode
Score
Rank/total
AI Agent - Tool Usage
4 evaluationsBenchmark / mode
Score
Rank/total
Compare with other models
Claude Fable 5.1
Publisher
Anthropic
View publisher details Claude Fable 5.1
Model Overview
Official release
Claude Fable 5.1 is Anthropic's generally available flagship model, released on September 1, 2026 under the API ID claude-fable-5-1. It is designed for demanding reasoning, long-horizon agentic coding, multistep research, and complex knowledge work. It shares an underlying model with invitation-only Claude Mythos 5.1 but uses safeguards intended for broad access.
Specifications and capabilities
Anthropic documents a 1M-token context window, 128K-token maximum output, text and image input, and text output. Reliable knowledge and training-data cutoffs are June 2026. Adaptive thinking is always on and default effort is high. Parameter count, training compute, and weight architecture are undisclosed.
Pricing and evaluations
Standard Claude API pricing per million tokens is $10 input and $50 output; cache reads cost $0.25 and batch input and output are 50% discounted. Anthropic reports 52.6% on Terminal-Bench-Science 0.1, 55.8% on Terminal-Bench 4.0, 1853 on GDPval-AA v2, 60.9% without tools and 65.0% with tools on Humanity's Last Exam, 31.4% on AutomationBench, and 73.4% on CursorBench 3.2.0.
Official sources
Claude Fable 5.1
FAQ
What are its context window and maximum output?
Anthropic documents a 1,000,000-token context window and a 128,000-token maximum output.
How much does the API cost?
Standard pricing is $10 per million input tokens and $50 per million output tokens. Cache reads cost $0.25, and batch input and output are 50% discounted.
Can adaptive thinking be disabled?
No. Adaptive thinking is always on; developers steer depth with effort, which defaults to high.
Is Claude Fable 5.1 open source?
No. Anthropic has not released weights or disclosed parameter count or training compute.
DataLearner on WeChat
Follow DataLearner on WeChat for AI model updates and research notes.
