Claude Fable 5
Explore benchmark scores, genre strengths, weaknesses, and recent examples for Claude Fable 5 on Orivel.
Model Overview
Released
2026-06-09
Context
1M tokens
Input
$10.00 / 1M
Output
$50.00 / 1M
Claude Fable 5 is Anthropic's most capable widely released model, generally available from June 9, 2026. It is the public release of the Mythos model family, positioned for the most demanding reasoning, long-horizon agentic work, vision, and scientific research.
In Anthropic's own testing, Fable 5 posts state-of-the-art results on nearly all measured benchmarks and beats not only the previous Anthropic flagship Opus 4.8 but also GPT-5.5 and Gemini 3.1 Pro. It ships with safety safeguards: in high-risk areas (cybersecurity, biology, chemistry) it blocks the response and falls back to Claude Opus 4.8 — these safeguards trigger in under 5% of sessions on average.
The model uses always-on adaptive thinking, the Opus 4.7-generation tokenizer, a 1M-token context window, and up to 128k tokens of output. Pricing is $10 input / $50 output per 1M tokens — double Opus 4.8 — reflecting its frontier tier. Knowledge cutoff is January 2026.
What changed
- Released June 9, 2026 as Anthropic's most capable widely released model (public release of the Mythos family)
- State-of-the-art on nearly all measured benchmarks; beats Opus 4.8, GPT-5.5, and Gemini 3.1 Pro in Anthropic's testing
- Exceptional at software engineering, knowledge work, vision, and scientific research
- Safety safeguards: blocks high-risk (cyber/bio/chem) responses and falls back to Opus 4.8; triggers in <5% of sessions on average
- Always-on adaptive thinking; Opus 4.7-generation tokenizer
- 1M-token context window; up to 128k output tokens
- Pricing: $10 input / $50 output per 1M tokens — double Opus 4.8 (frontier tier)
- Available across the Claude API, Amazon Bedrock, Vertex AI, and Microsoft Foundry
- Knowledge cutoff: January 2026
Overall Performance
Overall Rank
#1
Overall win rate
Average Score
Wins
3
Sample Count
3
Win Rate by Model
| Model | Wins | Losses | Draws | Win Rate | Detail |
|---|---|---|---|---|---|
| OpenAI GPT-5.4 | 1 | 0 | 0 |
100%
|
View Claude Fable 5 vs GPT-5.4 Comparison & Evaluation |
| Google Gemini 2.5 Flash | 1 | 0 | 0 |
100%
|
View Claude Fable 5 vs Gemini 2.5 Flash Comparison & Evaluation |
| Google Gemini 2.5 Flash-Lite | 1 | 0 | 0 |
100%
|
View Claude Fable 5 vs Gemini 2.5 Flash-Lite Comparison & Evaluation |
Compare by Genre
Strong Genres
Idea Generation
Average Score
Genre Average
Win Rate
Sample Count
1
Genre Rank
1 / 12
Wins
1
Discussion
Average Score
Genre Average
Win Rate
Sample Count
1
Genre Rank
2 / 13
Wins
1
Summarization
Average Score
Genre Average
Win Rate
Sample Count
1
Genre Rank
2 / 13
Wins
1
Weaker Genres
Strength by Evaluation Criteria
Average score by criterion (out of 10)
Coverage
Diversity
Faithfulness
Usefulness
Specificity
Structure
Clarity
Originality
Compression
Latest Tasks
Idea Generation
Low-Budget Ideas to Revitalize a Neighborhood Library
Generate 12 practical ideas for a small neighborhood public library that wants to attract more visitors over the next six months without spending much money. Th...
Summarization
Summarize Core Principles from 'The Art of War'
Summarize the following excerpt from Sun Tzu's 'The Art of War'. Your summary should be a single, coherent paragraph between 150 and 200 words. Focus on the cor...