Claude Fable 5.1

Claude model for demanding reasoning and long-horizon agentic work
- Context
- 1M
- Released
- 1mo ago
- Input / 1M tokens
- $10.00
- Output / 1M tokens
- $50.00
- Cached input / 1M tokens
- $0.25
- Cache write / 1M tokens
- $12.50
Sources:models.devOpenRouter
Benchmarks
DeepSWE is the primary coding benchmark where an exact model evaluation exists. Other sourced evaluations appear below; missing scores are not zero.
Not evaluated in DeepSWE 1.1 as of 2026-09-22. Results are only shown for exact model versions in the official dataset. Official leaderboard.
Other published benchmarks
| Evaluation | Reported score and source |
|---|---|
| ARC-AGI-2 | 88.8% Third-party evaluator · claude-fable-5-1_high Reasoning effort: high External evaluation via Epoch AIOriginal-source terms apply |
| ARC-AGI-2 | 78.3% Third-party evaluator · claude-fable-5-1_low Reasoning effort: low External evaluation via Epoch AIOriginal-source terms apply |
| ARC-AGI-2 | 90.0% Third-party evaluator · claude-fable-5-1_max Reasoning effort: max External evaluation via Epoch AIOriginal-source terms apply |
| ARC-AGI-2 | 86.3% Third-party evaluator · claude-fable-5-1_medium Reasoning effort: medium External evaluation via Epoch AIOriginal-source terms apply |
| ARC-AGI-2 | 90.0% Third-party evaluator · claude-fable-5-1_xhigh Reasoning effort: xhigh External evaluation via Epoch AIOriginal-source terms apply |
| FrontierMath-Tier-4-v2-Private | |
| FrontierMath-Tiers-1-3-v2-Private | |
| OTIS Mock AIME 2024-2025 | |
| SimpleQA Verified |
Benchmark compilation: Epoch AI, Capabilities & benchmarking (CC BY attribution). Scores are converted to percentages and model names matched to the Vertical Tabs catalog. Individual evaluations are credited to their original sources. External results may carry separate licences and evaluation configurations; do not interpret scores from different harnesses as directly comparable.
Specifications
- Developer
- Anthropic
- Model family
- Claude
- Type
- Language model
- Predecessor
- —
- Successor
- —
- Released
- September 1, 2026
- Model ID
anthropic/claude-fable-5.1- Openness
- Closed
- License
- —
- Input modalities
- text, image, file
- Output modalities
- text
- Modality
- text+image+file->text
- Tokenizer
- Claude
- Max output tokens
- 128K
- Knowledge cutoff
- 2026-06
- Hugging Face
- —
- Source freshness
- Metadata synced Oct 8, 2026, 10:17 PM UTC
Additional source evidence
- Source model
- anthropic/claude-fable-5-1
- Source family
- claude-fable
- Last updated
- 2026-09-01
- Knowledge cutoff
- 2026-06
- Model context
- 1M
- Model output limit
- 128K
- Open weights
- No
- File attachments
- Yes
- Reasoning
- Yes
- Tool calling
- Yes
- Temperature control
- No
Published benchmarks
| Benchmark | Score | Evaluation details | Evidence |
|---|---|---|---|
| Terminal-Bench-Science | 52.6 · score | production safeguards with fallback · 0.1 · 2026-09-01 | Source |
| Terminal-Bench | 55.8 · score | production safeguards with fallback · 4.0 · 2026-09-01 | Source |
| GDPval-AA | 1853 · Elo | production safeguards with fallback · 2 · 2026-09-01 | Source |
| OSWorld | 77.9 · partial score | August 2026 task release · production safeguards; interventions score zero; no fallback · 2.0 · 2026-09-01 | Source |
| OSWorld | 41.7 · strict success rate | August 2026 task release · production safeguards; interventions score zero; no fallback · 2.0 · 2026-09-01 | Source |
| Humanity's Last Exam | 60.9 · score | without tools; production safeguards with fallback · 2026-09-01 | Source |
| Humanity's Last Exam | 65 · score | with tools; production safeguards with fallback · 2026-09-01 | Source |
| AutomationBench | 31.4 · pass rate | production safeguards with fallback · 2026-09-01 | Source |
| CursorBench | 73.4 · score | max effort; production safeguards with fallback · 3.2.0 · 2026-09-01 | Source |
Complete source record (JSON)
{
"id": "anthropic/claude-fable-5-1",
"name": "Claude Fable 5.1",
"description": "Claude model for demanding reasoning and long-horizon agentic work",
"family": "claude-fable",
"last_updated": "2026-09-01",
"knowledge": "2026-06",
"attachment": true,
"reasoning": true,
"tool_call": true,
"temperature": false,
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"limit": {
"context": 1000000,
"output": 128000
},
"benchmarks": [
{
"name": "Terminal-Bench-Science",
"score": 52.6,
"source": "https://www.anthropic.com/claude-fable-and-mythos-5-1",
"metric": "score",
"variant": "production safeguards with fallback",
"version": "0.1",
"date": "2026-09-01"
},
{
"name": "Terminal-Bench",
"score": 55.8,
"source": "https://www.anthropic.com/claude-fable-and-mythos-5-1",
"metric": "score",
"variant": "production safeguards with fallback",
"version": "4.0",
"date": "2026-09-01"
},
{
"name": "GDPval-AA",
"score": 1853,
"source": "https://www.anthropic.com/claude-fable-and-mythos-5-1",
"metric": "Elo",
"variant": "production safeguards with fallback",
"version": "2",
"date": "2026-09-01"
},
{
"name": "OSWorld",
"score": 77.9,
"source": "https://www.anthropic.com/claude-fable-and-mythos-5-1",
"metric": "partial score",
"variant": "production safeguards; interventions score zero; no fallback",
"dataset": "August 2026 task release",
"version": "2.0",
"date": "2026-09-01"
},
{
"name": "OSWorld",
"score": 41.7,
"source": "https://www.anthropic.com/claude-fable-and-mythos-5-1",
"metric": "strict success rate",
"variant": "production safeguards; interventions score zero; no fallback",
"dataset": "August 2026 task release",
"version": "2.0",
"date": "2026-09-01"
},
{
"name": "Humanity's Last Exam",
"score": 60.9,
"source": "https://www.anthropic.com/claude-fable-and-mythos-5-1",
"metric": "score",
"variant": "without tools; production safeguards with fallback",
"date": "2026-09-01"
},
{
"name": "Humanity's Last Exam",
"score": 65,
"source": "https://www.anthropic.com/claude-fable-and-mythos-5-1",
"metric": "score",
"variant": "with tools; production safeguards with fallback",
"date": "2026-09-01"
},
{
"name": "AutomationBench",
"score": 31.4,
"source": "https://www.anthropic.com/claude-fable-and-mythos-5-1",
"metric": "pass rate",
"variant": "production safeguards with fallback",
"date": "2026-09-01"
},
{
"name": "CursorBench",
"score": 73.4,
"source": "https://www.anthropic.com/claude-fable-and-mythos-5-1",
"metric": "score",
"variant": "max effort; production safeguards with fallback",
"version": "3.2.0",
"date": "2026-09-01"
}
],
"release_date": "2026-09-01",
"open_weights": false
}Infrastructure
Provider availability
| Provider | Endpoint | Context | Input /M tokens | Output /M tokens | Cached input /M tokens | Cache write /M tokens | Uptime 1d | Data handling |
|---|---|---|---|---|---|---|---|---|
| amazon-bedrock | 1M | $10.00 | $50.00 | $0.25 | $12.50 | — | ZDR: NoTraining: Unknown Prompt logging: Unknown Retention: Unknown Checked 2026-10-15 | |
| amazon-bedrockBatch pricing | 1M | $10.00 | $50.00 | $0.25 | $12.50 | — | ZDR: NoTraining: Unknown Prompt logging: Unknown Retention: Unknown Checked 2026-10-15 | |
| anthropic | 1M | $10.00 | $50.00 | $0.25 | $12.50 | 100.00% | ZDR: NoTraining: Unknown Prompt logging: Unknown Retention: Unknown Checked 2026-10-15 | |
| anthropicBatch pricing | 1M | $10.00 | $50.00 | $0.25 | $12.50 | 100.00% | ZDR: NoTraining: Unknown Prompt logging: Unknown Retention: Unknown Checked 2026-10-15 | |
| azure | 1M | $10.00 | $50.00 | $0.25 | $12.50 | 100.00% | ZDR: NoTraining: Unknown Prompt logging: Unknown Retention: Unknown Checked 2026-10-15 | |
| azureBatch pricing | 1M | $10.00 | $50.00 | $0.25 | $12.50 | 100.00% | ZDR: NoTraining: Unknown Prompt logging: Unknown Retention: Unknown Checked 2026-10-15 | |
| google-vertex/global | 1M | $10.00 | $50.00 | $0.25 | $12.50 | 99.42% | ZDR: NoTraining: Unknown Prompt logging: Unknown Retention: Unknown Checked 2026-10-15 | |
| google-vertex/globalBatch pricing | 1M | $10.00 | $50.00 | $0.25 | $12.50 | 99.66% | ZDR: NoTraining: Unknown Prompt logging: Unknown Retention: Unknown Checked 2026-10-15 |
Provider data:OpenRouter
Text provenance
Watermarked · GlobalClaude text watermarking
Anthropic says newer Claude models include an invisible statistical watermark globally. It changes token selection rather than adding hidden characters.
Official explanation