Model comparison
Claude Fable 5.1 vs GPT-5.4
Claude Fable 5.1 is the stronger model overall, scoring 69.0 to 59.4 on the Noometry Index. GPT-5.4 costs 3.6× less per token, which makes it the better buy when Claude Fable 5.1's lead doesn't matter for your workload.
Last verified . 46 shared benchmarks.
Summary
- They share 46 benchmarks with published results for both. Claude Fable 5.1 scores higher in 9 categories and GPT-5.4 in 1 category; 10 gaps are clear of the uncertainty.
- The widest gap is in coding, where Claude Fable 5.1 leads 74.7 to 52.6.
- The biggest single-benchmark swing is MirrorCode: 73.3% for Claude Fable 5.1 and 15.6% for GPT-5.4.
- GPT-5.4 is cheaper at $2.50 / $15 per million input/output tokens, against $10 / $50 for Claude Fable 5.1.
- GPT-5.4 accepts more context: 1.05M tokens versus 1M.
Side by side
| Claude Fable 5.1 | GPT-5.4 | |
|---|---|---|
| Provider | Anthropic | OpenAI |
| Noometry Index | 69.0 | 59.4 |
| Released | 2026-09-01 | 2026-03-05 |
| Weights | Proprietary | Proprietary |
| Context window | 1M | 1.05M |
| Max output | 128K | 128K |
| Input $ / M tokens | $10 | $2.50 |
| Output $ / M tokens | $50 | $15 |
| Results tracked | 52 | 68 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Claude Fable 5.1 leads
Claude Fable 5.1: 74.7 (#1), GPT-5.4: 52.6 (#33)
| Benchmark | Claude Fable 5.1 | GPT-5.4 |
|---|---|---|
| LMArena WebDev | 1744 | 1465 |
| SciCode | 63.1% | 56.6% |
| GSO | 88.2% | 31.4% |
| WeirdML | 92.9% | 77.7% |
| LMArena Coding | 1528 | 1497 |
| MirrorCode | 73.3% | 15.6% |
| ALE-Bench | 2,143 | 1,607 |
| SWE-bench Verified | — | 76.9% |
| DeepSWE | — | 51.8% |
| FrontierCode | 50.9% | — |
| CursorBench | 51.8% | — |
| FrontierSWE | 56.3% | — |
| AlgoTune | — | 1.85 |
Agentic & Tool Use Claude Fable 5.1 leads
Claude Fable 5.1: 50.7 (#5), GPT-5.4: 46.5 (#13)
| Benchmark | Claude Fable 5.1 | GPT-5.4 |
|---|---|---|
| APEX-Agents | 68.6% | 52.4% |
| Vending-Bench 2 | 5,422 | 6,144 |
| Terminal-Bench | — | 81.8% |
| Remote Labor Index | 17.9% | — |
| τ²-bench Banking | — | 39.4% |
| DeepResearch Bench | — | 35.1% |
| PostTrainBench | — | 19% |
| GBAEval | — | 45.1% |
| GDP.pdf | 29.6% | — |
| LMArena Search | — | 1197 |
| METR Time Horizons | — | 74.3% |
Reasoning Claude Fable 5.1 leads
Claude Fable 5.1: 76.7 (#7), GPT-5.4: 61.8 (#19)
| Benchmark | Claude Fable 5.1 | GPT-5.4 |
|---|---|---|
| ARC-AGI-2 | 90% | 74% |
| NYT Connections (extended) | 90% | 91.3% |
| ARC-AGI-1 | 97.5% | 93.7% |
| CritPt | 31.1% | 23.4% |
| Chess Puzzles | 47% | 44% |
| EBR-Bench | 57.1% | 25.4% |
| LMArena Hard Prompts | 1526 | 1485 |
| Mystery Game Puzzles | 58% | 37% |
| DTBench | 97.6% | 94.4% |
| LMCA | 65.5% | 52% |
| Epoch Capabilities Index | 164.7 | 156.81 |
| Kagi LLM Benchmark | — | 63.8% |
| EnigmaEval | — | 16% |
| Thematic Generalization | — | 80% |
| ForecastBench | — | 59.5 |
Math Claude Fable 5.1 leads
Claude Fable 5.1: 89.6 (#4), GPT-5.4: 73.5 (#19)
| Benchmark | Claude Fable 5.1 | GPT-5.4 |
|---|---|---|
| FrontierMath (Tiers 1-3) | 90.2% | 78.6% |
| FrontierMath Tier 4 | 87.8% | 49% |
| OTIS Mock AIME 2024-2025 | 100% | 97.8% |
| ProofBench | 100% | 56% |
| LMArena Math | 1525 | 1488 |
| MathArena Final-Answer Competitions | — | 83.1% |
| FrontierMath (Feb 2025 set) | — | 47.6% |
| FrontierMath Erdős | 0% | — |
| FrontierMath Tier 4 (v1) | — | 27.1% |
Knowledge Claude Fable 5.1 leads
Claude Fable 5.1: 69.6 (#6), GPT-5.4: 65.3 (#14)
| Benchmark | Claude Fable 5.1 | GPT-5.4 |
|---|---|---|
| Humanity's Last Exam | 46.5% | 36.2% |
| SimpleQA Verified | 70.8% | 45.1% |
| LMArena Expert | 1535 | 1507 |
| GPQA Diamond | — | 93.3% |
| Vectara Hallucination Rate | — | 7% |
Multimodal Claude Fable 5.1 leads
Claude Fable 5.1: 53.9 (#4), GPT-5.4: 43.7 (#20)
| Benchmark | Claude Fable 5.1 | GPT-5.4 |
|---|---|---|
| LMArena Vision | 1318 | 1303 |
| Blueprint-Bench 2 | 41.9% | 27.1% |
| Furniture Assembly | 70% | 37.5% |
| LMArena Document | 1513 | 1471 |
Multilingual Claude Fable 5.1 leads
Claude Fable 5.1: 59.1 (#3), GPT-5.4: 56.2 (#23)
| Benchmark | Claude Fable 5.1 | GPT-5.4 |
|---|---|---|
| LMArena Non-English | 1507 | 1465 |
| LMArena Chinese | 1586 | 1519 |
| LMArena French | 1525 | 1493 |
| LMArena German | 1500 | 1472 |
| LMArena Japanese | 1543 | 1485 |
| LMArena Korean | 1534 | 1448 |
| LMArena Russian | 1521 | 1480 |
| LMArena Spanish | 1516 | 1454 |
Instruction Following Claude Fable 5.1 leads
Claude Fable 5.1: 79.2 (#6), GPT-5.4: 77.1 (#27)
| Benchmark | Claude Fable 5.1 | GPT-5.4 |
|---|---|---|
| LMArena Instruction Following | 1517 | 1469 |
Long Context GPT-5.4 leads
Claude Fable 5.1: 46.7 (#20), GPT-5.4: 50.3 (#8)
| Benchmark | Claude Fable 5.1 | GPT-5.4 |
|---|---|---|
| LMArena Longer Query | 1522 | 1473 |
| CL-bench | — | 27.9% |
| CL-bench Life | — | 21.7% |
Writing & Preference Claude Fable 5.1 leads
Claude Fable 5.1: 79.2 (#2), GPT-5.4: 71.9 (#17)
| Benchmark | Claude Fable 5.1 | GPT-5.4 |
|---|---|---|
| LMArena Text | 1510 | 1469 |
| LMArena Creative Writing | 1507 | 1439 |
| EQ-Bench Creative Writing | 2162 | 1840 |
| LMArena Multi-Turn | 1492 | 1482 |
| EQ-Bench 4 | — | 1272 |
Frequently asked questions
Is Claude Fable 5.1 better than GPT-5.4?
Claude Fable 5.1 is the stronger model overall, scoring 69.0 to 59.4 on the Noometry Index. GPT-5.4 costs 3.6× less per token, which makes it the better buy when Claude Fable 5.1's lead doesn't matter for your workload.
Which is cheaper, Claude Fable 5.1 or GPT-5.4?
GPT-5.4 is cheaper. It lists at $2.50 per million input tokens and $15 per million output tokens; Claude Fable 5.1 lists at $10 and $50.
Is Claude Fable 5.1 or GPT-5.4 better for coding?
Claude Fable 5.1 scores higher on coding benchmarks: 74.7 versus 52.6 in the Noometry coding category.
Which has the bigger context window?
GPT-5.4 does, with 1.05M tokens against 1M.
How many benchmarks do Claude Fable 5.1 and GPT-5.4 share?
46 benchmarks have published results for both models. Claude Fable 5.1 has 52 scored results on Noometry and GPT-5.4 has 68.