Model comparison
Claude Sonnet 4 vs GPT-5.6 Luna
GPT-5.6 Luna is the stronger model overall, scoring 54.6 to 40.8 on the Noometry Index.
Last verified . 32 shared benchmarks.
Summary
- They share 32 benchmarks with published results for both. Claude Sonnet 4 scores higher in 1 category and GPT-5.6 Luna in 9 categories; 10 gaps are clear of the uncertainty.
- The widest gap is in math, where GPT-5.6 Luna leads 77.7 to 43.3.
- The biggest single-benchmark swing is ARC-AGI-2: 5.9% for Claude Sonnet 4 and 59.5% for GPT-5.6 Luna.
- GPT-5.6 Luna is cheaper at $0.20 / $1.20 per million input/output tokens, against $3 / $15 for Claude Sonnet 4.
- GPT-5.6 Luna accepts more context: 1.05M tokens versus 200K.
Side by side
| Claude Sonnet 4 | GPT-5.6 Luna | |
|---|---|---|
| Provider | Anthropic | OpenAI |
| Noometry Index | 40.8 | 54.6 |
| Released | 2025-05-22 | 2026-07-09 |
| Weights | Proprietary | Proprietary |
| Context window | 200K | 1.05M |
| Max output | 64K | 128K |
| Input $ / M tokens | $3 | $0.20 |
| Output $ / M tokens | $15 | $1.20 |
| Results tracked | 58 | 52 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-5.6 Luna leads
Claude Sonnet 4: 43.5 (#88), GPT-5.6 Luna: 54.5 (#28)
| Benchmark | Claude Sonnet 4 | GPT-5.6 Luna |
|---|---|---|
| SciCode | 40% | 53.6% |
| WeirdML | 46.1% | 60.9% |
| LMArena Coding | 1414 | 1466 |
| ALE-Bench | 655.35 | 1,667 |
| DeepSWE | — | 67.2% |
| FrontierCode | — | 39.8% |
| SWE-bench Verified (bash only) | 64.9% | — |
| Aider Polyglot | 61.3% | — |
| CursorBench | — | 35.9% |
| LMArena WebDev | — | 1519 |
| GSO | 4.9% | — |
Agentic & Tool Use Claude Sonnet 4 leads
Claude Sonnet 4: 38.5 (#31), GPT-5.6 Luna: 34.4 (#45)
| Benchmark | Claude Sonnet 4 | GPT-5.6 Luna |
|---|---|---|
| APEX-Agents | — | 43% |
| TheAgentCompany | 33.1% | — |
| Cybench | 35% | — |
| DeepResearch Bench | 46.6% | — |
| OSWorld | 43.9% | — |
| BALROG | — | 45.6% |
| GDP.pdf | — | 22.7% |
| METR Time Horizons | 62% | — |
| Vending-Bench 2 | — | 4,095 |
Reasoning GPT-5.6 Luna leads
Claude Sonnet 4: 22.9 (#187), GPT-5.6 Luna: 47.6 (#43)
| Benchmark | Claude Sonnet 4 | GPT-5.6 Luna |
|---|---|---|
| ARC-AGI-2 | 5.9% | 59.5% |
| SimpleBench | 45.5% | 46.8% |
| Kagi LLM Benchmark | 73% | 49.1% |
| ARC-AGI-1 | 40% | 88% |
| CritPt | 0.3% | 20.6% |
| LMArena Hard Prompts | 1372 | 1451 |
| DTBench | 77.1% | 89.1% |
| LMCA | 29% | 48.5% |
| Epoch Capabilities Index | 141.69 | 156.39 |
| NYT Connections (extended) | — | 69.4% |
| Chess Puzzles | — | 40% |
| EnigmaEval | 3.1% | — |
| Mystery Game Puzzles | — | 21% |
| Surface Evolver Bench | — | 61.9% |
| ForecastBench | 60.2 | — |
Math GPT-5.6 Luna leads
Claude Sonnet 4: 43.3 (#80), GPT-5.6 Luna: 77.7 (#14)
| Benchmark | Claude Sonnet 4 | GPT-5.6 Luna |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 71.1% | 98.3% |
| LMArena Math | 1375 | 1458 |
| FrontierMath (Tiers 1-3) | — | 82.1% |
| FrontierMath Tier 4 | — | 61% |
| ProofBench | — | 60% |
| Omni-MATH | 60.2% | — |
| MATH Level 5 | 84.4% | — |
| FrontierMath (Feb 2025 set) | 4.1% | — |
| FrontierMath Tier 4 (v1) | 0% | — |
Knowledge GPT-5.6 Luna leads
Claude Sonnet 4: 41.8 (#108), GPT-5.6 Luna: 58.5 (#34)
| Benchmark | Claude Sonnet 4 | GPT-5.6 Luna |
|---|---|---|
| GPQA Diamond | 79.2% | 91.6% |
| LMArena Expert | 1372 | 1478 |
| Humanity's Last Exam | 7.8% | — |
| SimpleQA Verified | — | 41% |
| MMLU-Pro | 84.3% | — |
| Confabulations | 13.2% | — |
| Vectara Hallucination Rate | 10.3% | — |
| GPQA (HELM) | 70.6% | — |
Multimodal GPT-5.6 Luna leads
Claude Sonnet 4: 26.2 (#121), GPT-5.6 Luna: 42.7 (#28)
| Benchmark | Claude Sonnet 4 | GPT-5.6 Luna |
|---|---|---|
| LMArena Vision | 1191 | 1258 |
| GeoBench | 37% | — |
| VPCT | 34% | — |
| Blueprint-Bench 2 | — | 22.6% |
| Furniture Assembly | — | 42.5% |
| LMArena Document | — | 1457 |
| MindCube | 44.8% | — |
Multilingual GPT-5.6 Luna leads
Claude Sonnet 4: 46.7 (#156), GPT-5.6 Luna: 52.8 (#78)
| Benchmark | Claude Sonnet 4 | GPT-5.6 Luna |
|---|---|---|
| LMArena Non-English | 1333 | 1417 |
| LMArena Chinese | 1350 | 1470 |
| LMArena French | 1363 | 1456 |
| LMArena German | 1331 | 1454 |
| LMArena Japanese | 1302 | 1411 |
| LMArena Korean | 1291 | 1415 |
| LMArena Russian | 1355 | 1428 |
| LMArena Spanish | 1357 | 1448 |
Instruction Following GPT-5.6 Luna leads
Claude Sonnet 4: 71.7 (#145), GPT-5.6 Luna: 75.6 (#57)
| Benchmark | Claude Sonnet 4 | GPT-5.6 Luna |
|---|---|---|
| LMArena Instruction Following | 1376 | 1437 |
| IFEval | 84% | — |
Long Context GPT-5.6 Luna leads
Claude Sonnet 4: 33.7 (#259), GPT-5.6 Luna: 43.9 (#82)
| Benchmark | Claude Sonnet 4 | GPT-5.6 Luna |
|---|---|---|
| LMArena Longer Query | 1398 | 1436 |
| Fiction.LiveBench | 46.9% | — |
Writing & Preference GPT-5.6 Luna leads
Claude Sonnet 4: 57.1 (#132), GPT-5.6 Luna: 68.0 (#29)
| Benchmark | Claude Sonnet 4 | GPT-5.6 Luna |
|---|---|---|
| LMArena Text | 1351 | 1431 |
| LMArena Creative Writing | 1345 | 1396 |
| EQ-Bench Creative Writing | 1483 | 1829 |
| LMArena Multi-Turn | 1376 | 1434 |
| Short-Story Creative Writing | 81.4% | — |
| WildBench | 83.8% | — |
| EQ-Bench 4 | — | 1156 |
Frequently asked questions
Is Claude Sonnet 4 better than GPT-5.6 Luna?
GPT-5.6 Luna is the stronger model overall, scoring 54.6 to 40.8 on the Noometry Index.
Which is cheaper, Claude Sonnet 4 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 Sonnet 4 lists at $3 and $15.
Is Claude Sonnet 4 or GPT-5.6 Luna better for coding?
GPT-5.6 Luna scores higher on coding benchmarks: 54.5 versus 43.5 in the Noometry coding category.
Which has the bigger context window?
GPT-5.6 Luna does, with 1.05M tokens against 200K.
How many benchmarks do Claude Sonnet 4 and GPT-5.6 Luna share?
32 benchmarks have published results for both models. Claude Sonnet 4 has 58 scored results on Noometry and GPT-5.6 Luna has 52.