Model comparison
Claude Opus 4.6 vs GPT-5.4
GPT-5.4 is the stronger model overall, scoring 59.4 to 58.2 on the Noometry Index.
Last verified . 62 shared benchmarks.
Summary
- They share 62 benchmarks with published results for both. Claude Opus 4.6 scores higher in 5 categories and GPT-5.4 in 5 categories; 10 gaps are clear of the uncertainty.
- The widest gap is in math, where GPT-5.4 leads 73.5 to 63.0.
- The biggest single-benchmark swing is Chess Puzzles: 17% for Claude Opus 4.6 and 44% for GPT-5.4.
- GPT-5.4 is cheaper at $2.50 / $15 per million input/output tokens, against $5 / $25 for Claude Opus 4.6.
- GPT-5.4 accepts more context: 1.05M tokens versus 1M.
Side by side
| Claude Opus 4.6 | GPT-5.4 | |
|---|---|---|
| Provider | Anthropic | OpenAI |
| Noometry Index | 58.2 | 59.4 |
| Released | 2026-02-04 | 2026-03-05 |
| Weights | Proprietary | Proprietary |
| Context window | 1M | 1.05M |
| Max output | 128K | 128K |
| Input $ / M tokens | $5 | $2.50 |
| Output $ / M tokens | $25 | $15 |
| Results tracked | 68 | 68 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Claude Opus 4.6 leads
Claude Opus 4.6: 57.2 (#20), GPT-5.4: 52.6 (#33)
| Benchmark | Claude Opus 4.6 | GPT-5.4 |
|---|---|---|
| SWE-bench Verified | 78.7% | 76.9% |
| LMArena WebDev | 1547 | 1465 |
| GSO | 41.2% | 31.4% |
| WeirdML | 78% | 77.7% |
| LMArena Coding | 1536 | 1497 |
| ALE-Bench | 996.5 | 1,607 |
| AlgoTune | 1.47 | 1.85 |
| DeepSWE | — | 51.8% |
| FrontierCode | 26.6% | — |
| SWE-bench Verified (bash only) | 75.6% | — |
| SWE-bench Multilingual | 72% | — |
| SciCode | — | 56.6% |
| MirrorCode | — | 15.6% |
Agentic & Tool Use Claude Opus 4.6 leads
Claude Opus 4.6: 51.1 (#4), GPT-5.4: 46.5 (#13)
| Benchmark | Claude Opus 4.6 | GPT-5.4 |
|---|---|---|
| Terminal-Bench | 79.8% | 81.8% |
| APEX-Agents | 46.3% | 52.4% |
| τ²-bench Banking | 27.3% | 39.4% |
| DeepResearch Bench | 55.3% | 35.1% |
| GBAEval | 44.1% | 45.1% |
| LMArena Search | 1253 | 1197 |
| METR Time Horizons | 78.9% | 74.3% |
| Vending-Bench 2 | 8,018 | 6,144 |
| Remote Labor Index | 4.2% | — |
| Cybench | 93% | — |
| PostTrainBench | — | 19% |
Reasoning GPT-5.4 leads
Claude Opus 4.6: 57.8 (#23), GPT-5.4: 61.8 (#19)
| Benchmark | Claude Opus 4.6 | GPT-5.4 |
|---|---|---|
| ARC-AGI-2 | 69.2% | 74% |
| Kagi LLM Benchmark | 83.6% | 63.8% |
| NYT Connections (extended) | 92.1% | 91.3% |
| ARC-AGI-1 | 94% | 93.7% |
| Chess Puzzles | 17% | 44% |
| EnigmaEval | 7.6% | 16% |
| Thematic Generalization | 80.6% | 80% |
| EBR-Bench | 12.7% | 25.4% |
| LMArena Hard Prompts | 1527 | 1485 |
| Mystery Game Puzzles | 25% | 37% |
| DTBench | 91.2% | 94.4% |
| LMCA | 55.8% | 52% |
| Epoch Capabilities Index | 155.24 | 156.81 |
| ForecastBench | 60 | 59.5 |
| SimpleBench | 67.6% | — |
| CritPt | — | 23.4% |
Math GPT-5.4 leads
Claude Opus 4.6: 63.0 (#31), GPT-5.4: 73.5 (#19)
| Benchmark | Claude Opus 4.6 | GPT-5.4 |
|---|---|---|
| FrontierMath (Tiers 1-3) | 66% | 78.6% |
| FrontierMath Tier 4 | 26.8% | 49% |
| MathArena Final-Answer Competitions | 78.5% | 83.1% |
| OTIS Mock AIME 2024-2025 | 94.4% | 97.8% |
| ProofBench | 50% | 56% |
| LMArena Math | 1519 | 1488 |
| FrontierMath (Feb 2025 set) | 40.7% | 47.6% |
| FrontierMath Tier 4 (v1) | 22.9% | 27.1% |
Knowledge GPT-5.4 leads
Claude Opus 4.6: 61.9 (#26), GPT-5.4: 65.3 (#14)
| Benchmark | Claude Opus 4.6 | GPT-5.4 |
|---|---|---|
| GPQA Diamond | 90.5% | 93.3% |
| Humanity's Last Exam | 34.4% | 36.2% |
| SimpleQA Verified | 47% | 45.1% |
| Vectara Hallucination Rate | 12.2% | 7% |
| LMArena Expert | 1546 | 1507 |
Multimodal GPT-5.4 leads
Claude Opus 4.6: 37.3 (#74), GPT-5.4: 43.7 (#20)
| Benchmark | Claude Opus 4.6 | GPT-5.4 |
|---|---|---|
| LMArena Vision | 1316 | 1303 |
| Furniture Assembly | 28.3% | 37.5% |
| LMArena Document | 1507 | 1471 |
| Blueprint-Bench 2 | — | 27.1% |
Multilingual Claude Opus 4.6 leads
Claude Opus 4.6: 57.9 (#6), GPT-5.4: 56.2 (#23)
| Benchmark | Claude Opus 4.6 | GPT-5.4 |
|---|---|---|
| LMArena Non-English | 1489 | 1465 |
| LMArena Chinese | 1551 | 1519 |
| LMArena French | 1513 | 1493 |
| LMArena German | 1502 | 1472 |
| LMArena Japanese | 1484 | 1485 |
| LMArena Korean | 1464 | 1448 |
| LMArena Russian | 1497 | 1480 |
| LMArena Spanish | 1510 | 1454 |
Instruction Following Claude Opus 4.6 leads
Claude Opus 4.6: 79.5 (#4), GPT-5.4: 77.1 (#27)
| Benchmark | Claude Opus 4.6 | GPT-5.4 |
|---|---|---|
| LMArena Instruction Following | 1523 | 1469 |
Long Context GPT-5.4 leads
Claude Opus 4.6: 48.1 (#13), GPT-5.4: 50.3 (#8)
| Benchmark | Claude Opus 4.6 | GPT-5.4 |
|---|---|---|
| CL-bench | 20.7% | 27.9% |
| CL-bench Life | 17% | 21.7% |
| LMArena Longer Query | 1520 | 1473 |
Writing & Preference Claude Opus 4.6 leads
Claude Opus 4.6: 73.5 (#10), GPT-5.4: 71.9 (#17)
| Benchmark | Claude Opus 4.6 | GPT-5.4 |
|---|---|---|
| LMArena Text | 1503 | 1469 |
| LMArena Creative Writing | 1505 | 1439 |
| EQ-Bench Creative Writing | 1809 | 1840 |
| EQ-Bench 4 | 1223 | 1272 |
| LMArena Multi-Turn | 1513 | 1482 |
Frequently asked questions
Is Claude Opus 4.6 better than GPT-5.4?
GPT-5.4 is the stronger model overall, scoring 59.4 to 58.2 on the Noometry Index.
Which is cheaper, Claude Opus 4.6 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 Opus 4.6 lists at $5 and $25.
Is Claude Opus 4.6 or GPT-5.4 better for coding?
Claude Opus 4.6 scores higher on coding benchmarks: 57.2 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 Opus 4.6 and GPT-5.4 share?
62 benchmarks have published results for both models. Claude Opus 4.6 has 68 scored results on Noometry and GPT-5.4 has 68.