Claude Fable 5
Anthropic · Released 9 Jun 2026 · claude-fable-5
$10.00
Input price / 1M tokens
#59 of 61, cheapest first
$50.00
Output price / 1M tokens
#59 of 61, cheapest first
1M
Context window
#29 of 71
–
Output tokens / second
42%
Terminal-Bench 4.0
#7 of 10
72.9%
OSWorld 2.0
#2 of 10
57.8%
Humanity’s Last Exam (no tools)
#2 of 23
63.8%
Humanity’s Last Exam (with tools)
#5 of 16
Summary
Claude Fable 5 is a proprietary model from Anthropic, released on 9 Jun 2026.
At $10.00 input and $50.00 output per million tokens, it is #59 of 61 on input price, cheapest first.
Its 1M-token context window ranks #29 of 71.
Legacy model, superseded by Claude Fable 5.1. Built for demanding reasoning and long-horizon agentic work; Anthropic called it a Mythos-class model made safe for general use.
Benchmarks · 9 reported
| Benchmark | Score | Bar | Rank | Source |
|---|---|---|---|---|
| Terminal-Bench 4.0agentic coding; comparison column in a later post | 42% | #7 of 10 | Anthropic Fable 5.1 / Mythos 5.1 launch postSelf-reported | |
| Terminal-Bench-Science 0.1SE ±3.5–4.5 pts; comparison column in a later post | 24.7% | – | Anthropic Fable 5.1 / Mythos 5.1 launch postSelf-reported | |
| GDPval-AA v2knowledge work; comparison column in a later post | 1,723 | Elo-style | – | Anthropic Fable 5.1 / Mythos 5.1 launch postSelf-reported |
| OSWorld 2.0partial-credit score; August 2026 task release | 72.9% | #2 of 10 | Anthropic Fable 5.1 / Mythos 5.1 launch postSelf-reported | |
| OSWorld 2.0strict score; August 2026 task release | 36.1% | – | Anthropic Fable 5.1 / Mythos 5.1 launch postSelf-reported | |
| Humanity’s Last Exam (no tools)no tools | 57.8% | #2 of 23 | Anthropic Fable 5.1 / Mythos 5.1 launch postSelf-reported | |
| Humanity’s Last Exam (with tools)with tools | 63.8% | #5 of 16 | Anthropic Fable 5.1 / Mythos 5.1 launch postSelf-reported | |
| AutomationBenchbusiness workflows | 17.1% | #8 of 8 | Anthropic Fable 5.1 / Mythos 5.1 launch postSelf-reported | |
| CursorBench 3.2.0agentic coding | 70.5% | – | Anthropic Fable 5.1 / Mythos 5.1 launch postSelf-reported |
Self-reported means the lab ran the test itself. A rank appears only where several labs report the same version of a benchmark.
Similar models
Models with the most benchmarks in common with Claude Fable 5, and the closest scores.
Terminal-Bench 4.0
#7 of 10Long jobs in a real terminal
- Claude Fable 542
- Claude Fable 5.155.8
- Claude Opus 552.3
- GPT-6 Astra57.9
- Claude Opus 5.566.4
OSWorld 2.0
#2 of 10Long desktop and web workflows
- Claude Fable 572.9
- Claude Fable 5.177.9
- GPT-6 Astra72.6
Humanity’s Last Exam (no tools)
#2 of 23Expert questions, no search or code
- Claude Fable 557.8
- Claude Fable 5.160.9
- Claude Opus 556.6
- DeepSeek V4 Pro42.7
Humanity’s Last Exam (with tools)
#5 of 16Expert questions, with search and code
Compare Claude Fable 5 side by side
Pre-filled with the two closest models. Swap any of them.
Sources
- price, cache prices, batch, tokenizer: platform.claude.com/docs/en/about-claude/pricing, checked 2 Oct 2026
- context, max output, release, api id, knowledge cutoff, modalities, legacy status: platform.claude.com/docs/en/models/fable-5/overview, checked 2 Oct 2026
- status, retirement commitment (not sooner than 2027-06-09): platform.claude.com/docs/en/about-claude/model-deprecations, checked 2 Oct 2026
- classifiers, refusal billing change on 24 Sep 2026, data retention: platform.claude.com/docs/en/models/fable-5/introducing-claude-fable-5-and-claude, checked 2 Oct 2026
- June 2026 suspension and restoration: anthropic.com/news/redeploying-fable-5, checked 2 Oct 2026
- benchmarks (comparison column): anthropic.com/claude-fable-and-mythos-5-1, checked 2 Oct 2026
Building on Claude Fable 5?
Get the architecture right before the bill arrives.
We size caching, routing and fallbacks for your workload on a free call, and tell you where a cheaper model is good enough.