Claude Fable 5.1 vs GPT-5.4
Updated October 21, 2026. 91 benchmarks have a result for both. Coding is ranked on the Vibe Coding Index; every other board is shown as published.
Decision reading
Pick Claude Fable 5.1 for coding: it scores 89.1 on the Vibe Coding Index against 37.7 for GPT-5.4.
Pick GPT-5.4 when output tokens cost 3.3x less, or when the context window is larger, 1.05M against 1M tokens.
- Shared results
- 91
- Claude Fable 5.1 only
- 33
- GPT-5.4 only
- 18
- Like-for-like categories
- 8 of 9
Shared results can support a head-to-head reading. Results only one model has describe coverage, not superiority.
Which one for your work
Each card is decided by one number both models carry. Where only one has it, the card says so and makes no call.
Coding
Vibe Coding Index, four public coding benchmarks
Claude Fable 5.1
Claude Fable 5.1 scores 89.1 against 37.7.
2 and 1 benchmarks behind the scores
Intelligence
AA Intelligence Index
Claude Fable 5.1
Claude Fable 5.1 leads, 53.4 against 39.0.
Same board, both measured
Agentic work
AA Agentic Index
No call
Only Claude Fable 5.1 has a AA Agentic Index result, 57.9.
Directional only
Reasoning
GPQA Diamond
Claude Fable 5.1
Claude Fable 5.1 leads, 93.4% against 91.7%.
Same board, both measured
Math
AIME
No call
Only GPT-5.4 has a AIME result, 96.7%.
Directional only
Knowledge
MMLU Pro
Claude Fable 5.1
Claude Fable 5.1 leads, 92.4% against 87.5%.
Same board, both measured
Web design
Design Arena: websites
Claude Fable 5.1
Claude Fable 5.1 leads, 1313 against 1199.
Same board, both measured
Long documents
Listed context window
GPT-5.4
GPT-5.4 takes 1.05M tokens against 1M.
Listed by OpenRouter
Everyday cost
1K input + 500 output tokens
GPT-5.4
Claude Fable 5.1 $0.035, GPT-5.4 $0.010 per turn.
List prices
Cost per solved task
One SWE-bench Verified task, solved
No call
Only GPT-5.4 has the number, $1.02.
Directional only
At a glance
- Coding index
- Claude Fable 5.1 89.1 from 2 benchmarks, GPT-5.4 37.7 from 1 benchmark.
- Widest category
- Agentic work, 23.0 points apart on average over 6 shared benchmarks.
- Widest single gap
- MirrorCode: Claude Fable 5.1 73.3% against GPT-5.4 15.6%, 57.7 points apart on a 0 to 100 scale.
- Output price
- Claude Fable 5.1 $50.00 against GPT-5.4 $15.00 per million output tokens, 3.3x apart.
- Context window
- Claude Fable 5.1 takes 1M tokens, GPT-5.4 1.05M.
Benchmark by benchmark
First the score: boards each model leads, per category, where both have a result. Then every board, a category to a fold; bars are each result on a 0 to 100 scale, so compare the two in a row, not one row with another.
| Category | Claude Fable 5.1 | Bars | GPT-5.4 | Ties |
|---|---|---|---|---|
| Coding | 16 | 0 | ||
| Intelligence | 2 | 0 | ||
| Agentic work | 5 | 1 | ||
| Reasoning | 13 | 0 | 1 tie | |
| Math | 6 | 0 | ||
| Knowledge | 10 | 0 | ||
| Web design | 8 | 0 | ||
| General | 29 | 0 | ||
| All shared boards | 89 | 1 | 1 tie |
CodingLike-for-likeFour of these boards feed the Vibe Coding Index160
| Benchmark | Claude Fable 5.1 | GPT-5.4 | Leader |
|---|---|---|---|
| LiveBench coding | 97.7% | Not measured | Only Claude Fable 5.1 measured |
| LMArena WebDev | 1743 | 1464 | Claude Fable 5.1 |
| AA Coding Index | 81.6 | 71.1 | Claude Fable 5.1 |
| LiveCodeBench (Vals AI) | 90.5% | 84.1% | Claude Fable 5.1 |
| SWE-bench Verified (Vals AI) | Not measured | 78.2% | Only GPT-5.4 measured |
| Terminal-Bench 4 (Vals AI) | 58.1% | Not measured | Only Claude Fable 5.1 measured |
| Vibe Code Bench (Vals AI) | 90.3% | 67.4% | Claude Fable 5.1 |
| Code migration (Vals AI) | 54.6% | 35.0% | Claude Fable 5.1 |
| ProgramBench (Vals AI) | 7.0% | 0.5% | Claude Fable 5.1 |
| IOI (Vals AI) | 90.8% | Not measured | Only Claude Fable 5.1 measured |
| Terminal-Bench 2.1 (Vals AI) | 85.0% | Not measured | Only Claude Fable 5.1 measured |
| SWE-bench Multimodal (Vals AI) | 61.3% | Not measured | Only Claude Fable 5.1 measured |
| Vibe Code Bench 1-100 (Vals AI) | 28.0% | Not measured | Only Claude Fable 5.1 measured |
| SkillsBench (Vals AI) | 61.5% | 51.7% | Claude Fable 5.1 |
| CyberBench (Vals AI) | 70.4% | Not measured | Only Claude Fable 5.1 measured |
| SciCode | 63.1% | Not measured | Only Claude Fable 5.1 measured |
| Terminal-Bench Hard (Artificial Analysis run) | Not measured | 57.6% | Only GPT-5.4 measured |
| Terminal-Bench | Not measured | 81.8% | Only GPT-5.4 measured |
| FrontierCode | 50.9% | Not measured | Only Claude Fable 5.1 measured |
| FrontierSWE | 56.3% | Not measured | Only Claude Fable 5.1 measured |
| GSO-Bench | 88.2% | 31.4% | Claude Fable 5.1 |
| WeirdML | 92.9% | 77.7% | Claude Fable 5.1 |
| MirrorCode | 73.3% | 15.6% | Claude Fable 5.1 |
| ALE-Bench | 2143.2 | 1607.0 | Claude Fable 5.1 |
| LMArena Text: coding | 1532 | 1496 | Claude Fable 5.1 |
| LMArena Text: software and IT | 1530 | 1491 | Claude Fable 5.1 |
| LMArena WebDev: React | 1792 | 1454 | Claude Fable 5.1 |
| LMArena WebDev: HTML | 1684 | 1478 | Claude Fable 5.1 |
| LMArena WebDev: from an image | 1722 | 1436 | Claude Fable 5.1 |
| Neither model has a result on the other 9 boards here | |||
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index20
| Benchmark | Claude Fable 5.1 | GPT-5.4 | Leader |
|---|---|---|---|
| AA Intelligence Index | 53.4 | 39.0 | Claude Fable 5.1 |
| Epoch Capabilities Index | 166.5 | 157.3 | Claude Fable 5.1 |
Agentic workLike-for-likePublic boards, shown as published, kept out of the index51
| Benchmark | Claude Fable 5.1 | GPT-5.4 | Leader |
|---|---|---|---|
| AA Agentic Index | 57.9 | Not measured | Only Claude Fable 5.1 measured |
| Design Arena: agents building web apps | 1318 | 1059 | Claude Fable 5.1 |
| Design Arena: agents building full-stack apps | 1326 | 1017 | Claude Fable 5.1 |
| Design Arena: agents building mobile apps | 1267 | 1087 | Claude Fable 5.1 |
| Design Arena: agents building Android apps | Not measured | 1037 | Only GPT-5.4 measured |
| Design Arena: agents building Godot games | 1327 | 1135 | Claude Fable 5.1 |
| Design Arena: PowerPoint slides | 1328 | Not measured | Only Claude Fable 5.1 measured |
| Design Arena: spreadsheets | 1374 | Not measured | Only Claude Fable 5.1 measured |
| LiveBench research | 82.0% | Not measured | Only Claude Fable 5.1 measured |
| LiveBench browser | 100.0% | Not measured | Only Claude Fable 5.1 measured |
| LiveBench long agentic | 80.8% | Not measured | Only Claude Fable 5.1 measured |
| SREBench (Vals AI) | 22.9% | Not measured | Only Claude Fable 5.1 measured |
| Finance Agent (Vals AI) | 58.9% | Not measured | Only Claude Fable 5.1 measured |
| Tax Agent Bench (Vals AI) | 49.2% | Not measured | Only Claude Fable 5.1 measured |
| Legal Research Bench (Vals AI) | 55.3% | Not measured | Only Claude Fable 5.1 measured |
| Excel Modeling Benchmark (Vals AI) | 76.7% | Not measured | Only Claude Fable 5.1 measured |
| Terminal-Bench-Science (Vals AI) | 40.0% | Not measured | Only Claude Fable 5.1 measured |
| Time Horizon Index (Vals AI) | 63.3% | Not measured | Only Claude Fable 5.1 measured |
| CUA-bench (Vals AI) | 13.2% | Not measured | Only Claude Fable 5.1 measured |
| Robo-bench (Vals AI) | 5.8% | Not measured | Only Claude Fable 5.1 measured |
| τ²-Bench Telecom | Not measured | 87.1% | Only GPT-5.4 measured |
| APEX-Agents | 68.6% | 52.4% | Claude Fable 5.1 |
| METR time horizons | Not measured | 74.3% | Only GPT-5.4 measured |
| Remote Labor Index | 17.9% | Not measured | Only Claude Fable 5.1 measured |
| DeepResearch Bench | Not measured | 35.1% | Only GPT-5.4 measured |
| PostTrainBench | Not measured | 19.0% | Only GPT-5.4 measured |
| SpireClimb | 2.0% | Not measured | Only Claude Fable 5.1 measured |
| Vending-Bench 2 | $5,422 | $6,144 | GPT-5.4 |
| LMArena Search | 1246 | Not measured | Only Claude Fable 5.1 measured |
| Neither model has a result on the other 10 boards here | |||
ReasoningLike-for-likePublic boards, shown as published, kept out of the index130
| Benchmark | Claude Fable 5.1 | GPT-5.4 | Leader |
|---|---|---|---|
| GPQA Diamond | 93.4% | 91.7% | Claude Fable 5.1 |
| LiveBench reasoning | 91.7% | Not measured | Only Claude Fable 5.1 measured |
| MysteryMechanism (Vals AI) | 47.8% | Not measured | Only Claude Fable 5.1 measured |
| MMMU Pro (Vals AI) | 90.6% | 87.5% | Claude Fable 5.1 |
| GPQA Diamond (Artificial Analysis run) | 93.7% | 92.0% | Claude Fable 5.1 |
| Humanity's Last Exam (Artificial Analysis run) | 59.1% | 43.7% | Claude Fable 5.1 |
| GPQA Diamond (Epoch AI run) | Not measured | 93.3% | Only GPT-5.4 measured |
| ARC-AGI | 97.5% | 93.7% | Claude Fable 5.1 |
| ARC-AGI-2 | 90.0% | 74.0% | Claude Fable 5.1 |
| Chess puzzles | 48.0% | 48.0% | Tie |
| Mystery game puzzles | 58.0% | 37.0% | Claude Fable 5.1 |
| EBR-bench | 57.1% | 25.4% | Claude Fable 5.1 |
| LMArena Text: hard prompts | 1528 | 1485 | Claude Fable 5.1 |
| LMArena Text: expert | 1544 | 1507 | Claude Fable 5.1 |
| LMArena Text: hard prompts in English | 1522 | 1479 | Claude Fable 5.1 |
| LMArena Vision: diagrams | 1343 | 1327 | Claude Fable 5.1 |
| LMArena Vision: homework | 1345 | 1336 | Claude Fable 5.1 |
| Neither model has a result on the other 4 boards here | |||
MathLike-for-likePublic boards, shown as published, kept out of the index60
| Benchmark | Claude Fable 5.1 | GPT-5.4 | Leader |
|---|---|---|---|
| AIME | Not measured | 96.7% | Only GPT-5.4 measured |
| OTIS Mock AIME | 100.0% | 97.8% | Claude Fable 5.1 |
| FrontierMath, tiers 1 to 3 | 90.2% | 78.6% | Claude Fable 5.1 |
| FrontierMath, tier 4 | 87.8% | 49.0% | Claude Fable 5.1 |
| ProofBench | 100.0% | 56.0% | Claude Fable 5.1 |
| LMArena Text: math | 1531 | 1488 | Claude Fable 5.1 |
| LMArena Text: mathematical work | 1541 | 1497 | Claude Fable 5.1 |
| Neither model has a result on the other 6 boards here | |||
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index100
| Benchmark | Claude Fable 5.1 | GPT-5.4 | Leader |
|---|---|---|---|
| MMLU Pro | 92.4% | 87.5% | Claude Fable 5.1 |
| LegalBench (Vals AI) | 88.5% | 86.0% | Claude Fable 5.1 |
| CaseLaw (Vals AI) | Not measured | 63.8% | Only GPT-5.4 measured |
| CorpFin (Vals AI) | Not measured | 65.3% | Only GPT-5.4 measured |
| TaxEval (Vals AI) | 76.0% | 74.0% | Claude Fable 5.1 |
| MortgageTax (Vals AI) | 70.8% | 68.3% | Claude Fable 5.1 |
| MedQA (Vals AI) | Not measured | 96.1% | Only GPT-5.4 measured |
| MedCode (Vals AI) | 53.5% | 41.3% | Claude Fable 5.1 |
| SimpleQA Verified | 70.8% | 45.1% | Claude Fable 5.1 |
| LMArena Text: science | 1532 | 1479 | Claude Fable 5.1 |
| LMArena Text: medicine and healthcare | 1531 | 1458 | Claude Fable 5.1 |
| LMArena Text: legal and government | 1514 | 1486 | Claude Fable 5.1 |
| LMArena Text: business and finance | 1493 | 1474 | Claude Fable 5.1 |
| Neither model has a result on the other 4 boards here | |||
Web designLike-for-likePublic boards, shown as published, kept out of the index80
| Benchmark | Claude Fable 5.1 | GPT-5.4 | Leader |
|---|---|---|---|
| Design Arena: websites | 1313 | 1199 | Claude Fable 5.1 |
| Design Arena: UI components | 1338 | 1215 | Claude Fable 5.1 |
| Design Arena: game dev | 1382 | 1195 | Claude Fable 5.1 |
| Design Arena: data viz | 1344 | 1196 | Claude Fable 5.1 |
| Design Arena: 3D | 1393 | 1065 | Claude Fable 5.1 |
| Design Arena: all code categories | 1336 | 1180 | Claude Fable 5.1 |
| Design Arena: SVG | 1301 | 1141 | Claude Fable 5.1 |
| Design Arena: ASCII art | 1324 | 1188 | Claude Fable 5.1 |
| LiveBench design | 83.4% | Not measured | Only Claude Fable 5.1 measured |
GeneralLike-for-likePublic boards, shown as published, kept out of the index290
| Benchmark | Claude Fable 5.1 | GPT-5.4 | Leader |
|---|---|---|---|
| LiveBench data analysis | 80.3% | Not measured | Only Claude Fable 5.1 measured |
| Public Benefits Bench (Vals AI) | 74.9% | Not measured | Only Claude Fable 5.1 measured |
| SAGE (Vals AI) | 48.5% | 43.3% | Claude Fable 5.1 |
| MedScribe (Vals AI) | 91.3% | 77.5% | Claude Fable 5.1 |
| IFBench | Not measured | 73.9% | Only GPT-5.4 measured |
| Long context reasoning (AA-LCR) | 85.3% | 82.0% | Claude Fable 5.1 |
| DTBench | 97.6% | 94.4% | Claude Fable 5.1 |
| LMCA | 65.5% | 52.0% | Claude Fable 5.1 |
| CL-bench | Not measured | 27.9% | Only GPT-5.4 measured |
| Furniture assembly | 70.0% | 37.5% | Claude Fable 5.1 |
| CL-bench Life | Not measured | 21.7% | Only GPT-5.4 measured |
| LMArena Text | 1512 | 1469 | Claude Fable 5.1 |
| LMArena Text: instruction following | 1518 | 1469 | Claude Fable 5.1 |
| LMArena Text: creative writing | 1509 | 1439 | Claude Fable 5.1 |
| LMArena Vision | 1319 | 1303 | Claude Fable 5.1 |
| LMArena Text: English | 1506 | 1467 | Claude Fable 5.1 |
| LMArena Text: non-English | 1508 | 1465 | Claude Fable 5.1 |
| LMArena Text: Chinese | 1584 | 1520 | Claude Fable 5.1 |
| LMArena Text: German | 1499 | 1472 | Claude Fable 5.1 |
| LMArena Text: Japanese | 1562 | 1485 | Claude Fable 5.1 |
| LMArena Text: Korean | 1517 | 1449 | Claude Fable 5.1 |
| LMArena Text: French | 1523 | 1493 | Claude Fable 5.1 |
| LMArena Text: Spanish | 1520 | 1452 | Claude Fable 5.1 |
| LMArena Text: Russian | 1520 | 1480 | Claude Fable 5.1 |
| LMArena Text: Polish | 1525 | 1472 | Claude Fable 5.1 |
| LMArena Text: multi-turn | 1493 | 1482 | Claude Fable 5.1 |
| LMArena Text: longer queries | 1524 | 1473 | Claude Fable 5.1 |
| LMArena Text: writing and language | 1510 | 1458 | Claude Fable 5.1 |
| LMArena Text: entertainment and media | 1497 | 1438 | Claude Fable 5.1 |
| LMArena Vision: English | 1313 | 1301 | Claude Fable 5.1 |
| LMArena Vision: Chinese | 1391 | 1354 | Claude Fable 5.1 |
| LMArena Vision: reading text | 1327 | 1312 | Claude Fable 5.1 |
| LMArena Vision: humor | 1318 | 1284 | Claude Fable 5.1 |
| LMArena Vision: recognizing things | Not measured | 1251 | Only GPT-5.4 measured |
| LMArena Vision: creative writing | 1330 | 1283 | Claude Fable 5.1 |
| LMArena Document | Not measured | 1471 | Only GPT-5.4 measured |
| Neither model has a result on the other 10 boards here | |||
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
| Benchmark | Claude Fable 5.1 | GPT-5.4 | Leader |
|---|---|---|---|
| Neither model has a result on 5 boards here | |||
What each workload costs
Three fixed token mixes at each model's listed price, so per-token rates become a bill.
Chat turn
1K input + 500 output tokens
GPT-5.4 costs less
Repository review
50K input + 3K output tokens
GPT-5.4 costs less
Agent loop
200K input + 10K output tokens
GPT-5.4 costs less
Listed OpenRouter rates, without cache or batch discounts. A workload that does not fit a model's context window would need splitting.
Specification differences
The leading value in a row is marked; a row with a missing value names no leader. Our two rows are marked in red.
| Specification | Anthropic Claude Fable 5.1 | OpenAI GPT-5.4 |
|---|---|---|
| Maker | Anthropic | OpenAI |
| Weights | No public weights listed | No public weights listed |
| Released | September 1, 2026 | March 5, 2026 |
| Context window | 1M | 1.05M |
| Input $/Mtok | $10.00 | $2.50 |
| Output $/Mtok | $50.00 | $15.00 |
| Coding index | 89.1 | 37.7 |
| Independent measurements behind it | 2 | 1 |
| Cost per solved task | Not computable | $1.02 |
| Source | OpenRouter listing | OpenRouter listing |
Questions
Is Claude Fable 5.1 better than GPT-5.4 for coding?
Pick Claude Fable 5.1 for coding: it scores 89.1 on the Vibe Coding Index against 37.7 for GPT-5.4. Pick GPT-5.4 when output tokens cost 3.3x less, or when the context window is larger, 1.05M against 1M tokens.
Which is cheaper, Claude Fable 5.1 or GPT-5.4?
Claude Fable 5.1 costs $10.00 per million input tokens and $50.00 per million output tokens; GPT-5.4 costs $2.50 and $15.00. Cost per solved task is known for GPT-5.4 only.
Which has the bigger context window, Claude Fable 5.1 or GPT-5.4?
GPT-5.4, with 1.05M tokens against 1M.
How many benchmarks is the Claude Fable 5.1 vs GPT-5.4 comparison based on?
91 benchmarks have a result for both models as of October 21, 2026. Claude Fable 5.1's coding index rests on 2 independent benchmarks and GPT-5.4's on 1. Where only one model has a result, the page says so instead of naming a winner.