# Claude Fable 5 vs DeepSeek-R1

> Claude Fable 5 is the stronger model overall, scoring 66.8 to 42.3 on the Noometry Index. DeepSeek-R1 costs 22× less per token, which makes it the better buy when Claude Fable 5's lead doesn't matter for your workload.

- Canonical page: https://noometry.com/compare/claude-fable-5-vs-deepseek-r1
- Last updated: 2026-10-10
- Shared benchmarks: 29

## Summary

- They share 29 benchmarks with published results for both. Claude Fable 5 scores higher in 9 categories and DeepSeek-R1 in 0 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where Claude Fable 5 leads 76.8 to 18.6.
- The biggest single-benchmark swing is ARC-AGI-2: 89.2% for Claude Fable 5 and 1.3% for DeepSeek-R1.
- DeepSeek-R1 is cheaper at $0.50 / $2.15 per million input/output tokens, against $10 / $50 for Claude Fable 5.
- Claude Fable 5 accepts more context: 1M tokens versus 164K.

## Snapshot

| | Claude Fable 5 | DeepSeek-R1 |
|---|---|---|
| Provider | Anthropic | DeepSeek |
| Noometry Index | 66.8 | 42.3 |
| Rank | 5 | 115 |
| Context | 1M | 164K |
| Input $/M | $10 | $0.50 |
| Output $/M | $50 | $2.15 |
| Weights | Proprietary | Proprietary |

## Coding

- Claude Fable 5: 70.6 (#4)
- DeepSeek-R1: 46.3 (#68)

| Benchmark | Claude Fable 5 | DeepSeek-R1 |
|---|---|---|
| SciCode | 61% | 35.7% |
| WeirdML | 91.9% | 41.6% |
| LMArena Coding | 1519 | 1427 |
| ALE-Bench | 2,041 | 804.12 |
| DeepSWE | 69.9% | — |
| FrontierCode | 53.5% | — |
| Aider Polyglot | — | 71.4% |
| LMArena WebDev | 1625 | — |
| FrontierSWE | 47% | — |
| GSO | 78.4% | — |
| LiveBench Coding | — | 66.7% |
| MirrorCode | 63.9% | — |
| AlgoTune | — | 1.7 |

## Agentic & Tool Use

- Claude Fable 5: 54.0 (#2)
- DeepSeek-R1: 30.7 (#75)

| Benchmark | Claude Fable 5 | DeepSeek-R1 |
|---|---|---|
| APEX-Agents | 63.6% | — |
| Remote Labor Index | 16.1% | — |
| τ²-bench Banking | 39.7% | — |
| DeepResearch Bench | — | 35.1% |
| PostTrainBench | 41.8% | — |
| BALROG | — | 34.9% |
| GBAEval | 74.5% | — |
| GDP.pdf | 30% | — |
| LMArena Search | 1230 | — |
| METR Time Horizons | — | 53.8% |
| Vending-Bench 2 | 5,680 | — |

## Reasoning

- Claude Fable 5: 76.8 (#6)
- DeepSeek-R1: 18.6 (#278)

| Benchmark | Claude Fable 5 | DeepSeek-R1 |
|---|---|---|
| ARC-AGI-2 | 89.2% | 1.3% |
| SimpleBench | 81.9% | 40.8% |
| Kagi LLM Benchmark | 91.4% | 69.4% |
| ARC-AGI-1 | 98.5% | 21.2% |
| CritPt | 28.6% | 1.1% |
| LMArena Hard Prompts | 1508 | 1416 |
| Epoch Capabilities Index | 162.06 | 141.29 |
| NYT Connections (extended) | 92.7% | — |
| Chess Puzzles | 41% | — |
| EnigmaEval | 39.3% | — |
| EBR-Bench | 39.5% | — |
| LiveBench Reasoning | — | 83.2% |
| Mystery Game Puzzles | 52% | — |
| DTBench | 98.4% | — |
| LiveBench Data Analysis | — | 69.8% |
| LMCA | 61.1% | — |
| Surface Evolver Bench | 95% | — |
| Bench to the Future 3 | 0.13 | — |
| ForecastBench | — | 60 |
| LiveBench | — | 71.6% |

## Math

- Claude Fable 5: 88.5 (#5)
- DeepSeek-R1: 43.8 (#79)

| Benchmark | Claude Fable 5 | DeepSeek-R1 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 100% | 66.4% |
| LMArena Math | 1519 | 1400 |
| FrontierMath (Tiers 1-3) | 87% | — |
| FrontierMath Tier 4 | 90.2% | — |
| ProofBench | 95% | — |
| Omni-MATH | — | 42.4% |
| LiveBench Math | — | 80.7% |
| MATH Level 5 | — | 96.6% |
| FrontierMath Erdős | 0% | — |

## Knowledge

- Claude Fable 5: 62.2 (#25)
- DeepSeek-R1: 44.5 (#87)

| Benchmark | Claude Fable 5 | DeepSeek-R1 |
|---|---|---|
| GPQA Diamond | 85.9% | 76.3% |
| LMArena Expert | 1534 | 1394 |
| SimpleQA Verified | 70.7% | — |
| MMLU-Pro | — | 79.3% |
| Confabulations | — | 12.7% |
| Vectara Hallucination Rate | — | 11.3% |
| GPQA (HELM) | — | 66.6% |

## Multimodal

- Claude Fable 5: 45.3 (#17)
- DeepSeek-R1: —

| Benchmark | Claude Fable 5 | DeepSeek-R1 |
|---|---|---|
| LMArena Vision | 1324 | — |
| Blueprint-Bench 2 | 38.6% | — |
| Furniture Assembly | 35.8% | — |
| LMArena Document | 1496 | — |

## Multilingual

- Claude Fable 5: 57.3 (#9)
- DeepSeek-R1: 52.4 (#85)

| Benchmark | Claude Fable 5 | DeepSeek-R1 |
|---|---|---|
| LMArena Non-English | 1481 | 1412 |
| LMArena Chinese | 1543 | 1442 |
| LMArena French | 1505 | 1417 |
| LMArena German | 1486 | 1404 |
| LMArena Japanese | 1506 | 1391 |
| LMArena Korean | 1488 | 1360 |
| LMArena Russian | 1504 | 1423 |
| LMArena Spanish | 1498 | 1411 |

## Instruction Following

- Claude Fable 5: 78.6 (#8)
- DeepSeek-R1: 72.0 (#143)

| Benchmark | Claude Fable 5 | DeepSeek-R1 |
|---|---|---|
| LMArena Instruction Following | 1502 | 1382 |
| LiveBench Instruction Following | — | 80.5% |
| IFEval | — | 78.4% |

## Long Context

- Claude Fable 5: 46.3 (#23)
- DeepSeek-R1: 45.4 (#36)

| Benchmark | Claude Fable 5 | DeepSeek-R1 |
|---|---|---|
| LMArena Longer Query | 1509 | 1391 |
| Fiction.LiveBench | — | 75% |

## Writing & Preference

- Claude Fable 5: 75.9 (#5)
- DeepSeek-R1: 61.4 (#88)

| Benchmark | Claude Fable 5 | DeepSeek-R1 |
|---|---|---|
| LMArena Text | 1491 | 1428 |
| LMArena Creative Writing | 1494 | 1405 |
| EQ-Bench Creative Writing | 1943 | 1500 |
| LMArena Multi-Turn | 1504 | 1405 |
| Short-Story Creative Writing | — | 83% |
| WildBench | — | 82.8% |
| EQ-Bench 4 | 1340 | — |
| LiveBench Language | — | 48.5% |

## FAQ

### Is Claude Fable 5 better than DeepSeek-R1?

Claude Fable 5 is the stronger model overall, scoring 66.8 to 42.3 on the Noometry Index. DeepSeek-R1 costs 22× less per token, which makes it the better buy when Claude Fable 5's lead doesn't matter for your workload.

### Which is cheaper, Claude Fable 5 or DeepSeek-R1?

DeepSeek-R1 is cheaper. It lists at $0.50 per million input tokens and $2.15 per million output tokens; Claude Fable 5 lists at $10 and $50.

### Is Claude Fable 5 or DeepSeek-R1 better for coding?

Claude Fable 5 scores higher on coding benchmarks: 70.6 versus 46.3 in the Noometry coding category.

### Which has the bigger context window?

Claude Fable 5 does, with 1M tokens against 164K.

### How many benchmarks do Claude Fable 5 and DeepSeek-R1 share?

29 benchmarks have published results for both models. Claude Fable 5 has 62 scored results on Noometry and DeepSeek-R1 has 52.