Model comparison
Claude Fable 5 vs GPT-5.6 Luna
Claude Fable 5 is the stronger model overall, scoring 66.8 to 54.6 on the Noometry Index. GPT-5.6 Luna costs 44× less per token, which makes it the better buy when Claude Fable 5's lead doesn't matter for your workload.
Last verified . 50 shared benchmarks.
Summary
- They share 50 benchmarks with published results for both. Claude Fable 5 scores higher in 10 categories and GPT-5.6 Luna in 0 categories; 10 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where Claude Fable 5 leads 76.8 to 47.6.
- The biggest single-benchmark swing is Kagi LLM Benchmark: 91.4% for Claude Fable 5 and 49.1% for GPT-5.6 Luna.
- GPT-5.6 Luna is cheaper at $0.20 / $1.20 per million input/output tokens, against $10 / $50 for Claude Fable 5.
- GPT-5.6 Luna accepts more context: 1.05M tokens versus 1M.
Side by side
| Claude Fable 5 | GPT-5.6 Luna | |
|---|---|---|
| Provider | Anthropic | OpenAI |
| Noometry Index | 66.8 | 54.6 |
| Released | 2026-06-07 | 2026-07-09 |
| Weights | Proprietary | Proprietary |
| Context window | 1M | 1.05M |
| Max output | 128K | 128K |
| Input $ / M tokens | $10 | $0.20 |
| Output $ / M tokens | $50 | $1.20 |
| Results tracked | 62 | 52 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Claude Fable 5 leads
Claude Fable 5: 70.6 (#4), GPT-5.6 Luna: 54.5 (#28)
| Benchmark | Claude Fable 5 | GPT-5.6 Luna |
|---|---|---|
| DeepSWE | 69.9% | 67.2% |
| FrontierCode | 53.5% | 39.8% |
| LMArena WebDev | 1625 | 1519 |
| SciCode | 61% | 53.6% |
| WeirdML | 91.9% | 60.9% |
| LMArena Coding | 1519 | 1466 |
| ALE-Bench | 2,041 | 1,667 |
| CursorBench | — | 35.9% |
| FrontierSWE | 47% | — |
| GSO | 78.4% | — |
| MirrorCode | 63.9% | — |
Agentic & Tool Use Claude Fable 5 leads
Claude Fable 5: 54.0 (#2), GPT-5.6 Luna: 34.4 (#45)
| Benchmark | Claude Fable 5 | GPT-5.6 Luna |
|---|---|---|
| APEX-Agents | 63.6% | 43% |
| GDP.pdf | 30% | 22.7% |
| Vending-Bench 2 | 5,680 | 4,095 |
| Remote Labor Index | 16.1% | — |
| τ²-bench Banking | 39.7% | — |
| PostTrainBench | 41.8% | — |
| BALROG | — | 45.6% |
| GBAEval | 74.5% | — |
| LMArena Search | 1230 | — |
Reasoning Claude Fable 5 leads
Claude Fable 5: 76.8 (#6), GPT-5.6 Luna: 47.6 (#43)
| Benchmark | Claude Fable 5 | GPT-5.6 Luna |
|---|---|---|
| ARC-AGI-2 | 89.2% | 59.5% |
| SimpleBench | 81.9% | 46.8% |
| Kagi LLM Benchmark | 91.4% | 49.1% |
| NYT Connections (extended) | 92.7% | 69.4% |
| ARC-AGI-1 | 98.5% | 88% |
| CritPt | 28.6% | 20.6% |
| Chess Puzzles | 41% | 40% |
| LMArena Hard Prompts | 1508 | 1451 |
| Mystery Game Puzzles | 52% | 21% |
| DTBench | 98.4% | 89.1% |
| LMCA | 61.1% | 48.5% |
| Surface Evolver Bench | 95% | 61.9% |
| Epoch Capabilities Index | 162.06 | 156.39 |
| EnigmaEval | 39.3% | — |
| EBR-Bench | 39.5% | — |
| Bench to the Future 3 | 0.13 | — |
Math Claude Fable 5 leads
Claude Fable 5: 88.5 (#5), GPT-5.6 Luna: 77.7 (#14)
| Benchmark | Claude Fable 5 | GPT-5.6 Luna |
|---|---|---|
| FrontierMath (Tiers 1-3) | 87% | 82.1% |
| FrontierMath Tier 4 | 90.2% | 61% |
| OTIS Mock AIME 2024-2025 | 100% | 98.3% |
| ProofBench | 95% | 60% |
| LMArena Math | 1519 | 1458 |
| FrontierMath Erdős | 0% | — |
Knowledge Claude Fable 5 leads
Claude Fable 5: 62.2 (#25), GPT-5.6 Luna: 58.5 (#34)
| Benchmark | Claude Fable 5 | GPT-5.6 Luna |
|---|---|---|
| GPQA Diamond | 85.9% | 91.6% |
| SimpleQA Verified | 70.7% | 41% |
| LMArena Expert | 1534 | 1478 |
Multimodal Claude Fable 5 leads
Claude Fable 5: 45.3 (#17), GPT-5.6 Luna: 42.7 (#28)
| Benchmark | Claude Fable 5 | GPT-5.6 Luna |
|---|---|---|
| LMArena Vision | 1324 | 1258 |
| Blueprint-Bench 2 | 38.6% | 22.6% |
| Furniture Assembly | 35.8% | 42.5% |
| LMArena Document | 1496 | 1457 |
Multilingual Claude Fable 5 leads
Claude Fable 5: 57.3 (#9), GPT-5.6 Luna: 52.8 (#78)
| Benchmark | Claude Fable 5 | GPT-5.6 Luna |
|---|---|---|
| LMArena Non-English | 1481 | 1417 |
| LMArena Chinese | 1543 | 1470 |
| LMArena French | 1505 | 1456 |
| LMArena German | 1486 | 1454 |
| LMArena Japanese | 1506 | 1411 |
| LMArena Korean | 1488 | 1415 |
| LMArena Russian | 1504 | 1428 |
| LMArena Spanish | 1498 | 1448 |
Instruction Following Claude Fable 5 leads
Claude Fable 5: 78.6 (#8), GPT-5.6 Luna: 75.6 (#57)
| Benchmark | Claude Fable 5 | GPT-5.6 Luna |
|---|---|---|
| LMArena Instruction Following | 1502 | 1437 |
Long Context Claude Fable 5 leads
Claude Fable 5: 46.3 (#23), GPT-5.6 Luna: 43.9 (#82)
| Benchmark | Claude Fable 5 | GPT-5.6 Luna |
|---|---|---|
| LMArena Longer Query | 1509 | 1436 |
Writing & Preference Claude Fable 5 leads
Claude Fable 5: 75.9 (#5), GPT-5.6 Luna: 68.0 (#29)
| Benchmark | Claude Fable 5 | GPT-5.6 Luna |
|---|---|---|
| LMArena Text | 1491 | 1431 |
| LMArena Creative Writing | 1494 | 1396 |
| EQ-Bench Creative Writing | 1943 | 1829 |
| EQ-Bench 4 | 1340 | 1156 |
| LMArena Multi-Turn | 1504 | 1434 |
Frequently asked questions
Is Claude Fable 5 better than GPT-5.6 Luna?
Claude Fable 5 is the stronger model overall, scoring 66.8 to 54.6 on the Noometry Index. GPT-5.6 Luna costs 44× 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 GPT-5.6 Luna?
GPT-5.6 Luna is cheaper. It lists at $0.20 per million input tokens and $1.20 per million output tokens; Claude Fable 5 lists at $10 and $50.
Is Claude Fable 5 or GPT-5.6 Luna better for coding?
Claude Fable 5 scores higher on coding benchmarks: 70.6 versus 54.5 in the Noometry coding category.
Which has the bigger context window?
GPT-5.6 Luna does, with 1.05M tokens against 1M.
How many benchmarks do Claude Fable 5 and GPT-5.6 Luna share?
50 benchmarks have published results for both models. Claude Fable 5 has 62 scored results on Noometry and GPT-5.6 Luna has 52.