Model comparison
GPT-5.4 vs Qwen3.6 Max Preview
GPT-5.4 is the stronger model overall, scoring 59.5 to 51.1 on the Noometry Index. Qwen3.6 Max Preview costs 1.9× less per token, which makes it the better buy when GPT-5.4's lead doesn't matter for your workload.
Last verified . 28 shared benchmarks.
Alibaba (Qwen)Summary
- They share 28 benchmarks with published results for both. GPT-5.4 scores higher in 8 categories and Qwen3.6 Max Preview in 0 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in math, where GPT-5.4 leads 73.3 to 52.2.
- The biggest single-benchmark swing is Chess Puzzles: 44% for GPT-5.4 and 20% for Qwen3.6 Max Preview.
- Qwen3.6 Max Preview is cheaper at $1.30 / $7.80 per million input/output tokens, against $2.50 / $15 for GPT-5.4.
- GPT-5.4 accepts more context: 1.05M tokens versus 262K.
Side by side
| GPT-5.4 | Qwen3.6 Max Preview | |
|---|---|---|
| Provider | OpenAI | Alibaba (Qwen) |
| Noometry Index | 59.5 | 51.1 |
| Released | 2026-03-05 | 2026-04-20 |
| Weights | Proprietary | Proprietary |
| Context window | 1.05M | 262K |
| Max output | 128K | 66K |
| Input $ / M tokens | $2.50 | $1.30 |
| Output $ / M tokens | $15 | $7.80 |
| Results tracked | 68 | 29 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-5.4 leads
GPT-5.4: 52.8 (#28), Qwen3.6 Max Preview: 48.6 (#50)
| Benchmark | GPT-5.4 | Qwen3.6 Max Preview |
|---|---|---|
| SWE-bench Verified | 76.9% | 76.7% |
| LMArena WebDev | 1465 | 1482 |
| LMArena Coding | 1497 | 1472 |
| DeepSWE | 51.8% | — |
| SciCode | 56.6% | — |
| GSO | 31.4% | — |
| WeirdML | 77.7% | — |
| MirrorCode | 15.6% | — |
| ALE-Bench | 1,607 | — |
| AlgoTune | 1.85 | — |
Agentic & Tool Use Not comparable
GPT-5.4: 46.5 (#12), Qwen3.6 Max Preview: —
| Benchmark | GPT-5.4 | Qwen3.6 Max Preview |
|---|---|---|
| Vending-Bench 2 | 6,144 | 4,254 |
| Terminal-Bench | 81.8% | — |
| APEX-Agents | 52.4% | — |
| τ²-bench Banking | 39.4% | — |
| DeepResearch Bench | 35.1% | — |
| PostTrainBench | 19% | — |
| GBAEval | 45.1% | — |
| LMArena Search | 1197 | — |
| METR Time Horizons | 74.3% | — |
Reasoning GPT-5.4 leads
GPT-5.4: 61.7 (#17), Qwen3.6 Max Preview: 40.8 (#47)
| Benchmark | GPT-5.4 | Qwen3.6 Max Preview |
|---|---|---|
| NYT Connections (extended) | 91.3% | 74.1% |
| Chess Puzzles | 44% | 20% |
| LMArena Hard Prompts | 1485 | 1457 |
| Mystery Game Puzzles | 37% | 19% |
| DTBench | 94.4% | 87.2% |
| LMCA | 52% | 42.5% |
| Epoch Capabilities Index | 156.92 | 149.27 |
| ARC-AGI-2 | 74% | — |
| SimpleBench | — | 63% |
| Kagi LLM Benchmark | 63.8% | — |
| ARC-AGI-1 | 93.7% | — |
| CritPt | 23.4% | — |
| EnigmaEval | 16% | — |
| Thematic Generalization | 80% | — |
| EBR-Bench | 25.4% | — |
| ForecastBench | 59.5 | — |
Math GPT-5.4 leads
GPT-5.4: 73.3 (#14), Qwen3.6 Max Preview: 52.2 (#40)
| Benchmark | GPT-5.4 | Qwen3.6 Max Preview |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 97.8% | 91.1% |
| LMArena Math | 1488 | 1465 |
| FrontierMath (Feb 2025 set) | 47.6% | 23.1% |
| FrontierMath Tier 4 (v1) | 27.1% | 4.2% |
| FrontierMath (Tiers 1-3) | 78.6% | — |
| FrontierMath Tier 4 | 49% | — |
| MathArena Final-Answer Competitions | 83.1% | — |
| ProofBench | 56% | — |
Knowledge GPT-5.4 leads
GPT-5.4: 65.3 (#13), Qwen3.6 Max Preview: 57.6 (#35)
| Benchmark | GPT-5.4 | Qwen3.6 Max Preview |
|---|---|---|
| GPQA Diamond | 93.3% | 87.4% |
| SimpleQA Verified | 45.1% | 52% |
| LMArena Expert | 1507 | 1476 |
| Humanity's Last Exam | 36.2% | — |
| Vectara Hallucination Rate | 7% | — |
Multimodal Not comparable
GPT-5.4: 43.8 (#19), Qwen3.6 Max Preview: —
| Benchmark | GPT-5.4 | Qwen3.6 Max Preview |
|---|---|---|
| LMArena Vision | 1302 | — |
| Blueprint-Bench 2 | 27.1% | — |
| Furniture Assembly | 37.5% | — |
| LMArena Document | 1471 | — |
Multilingual GPT-5.4 leads
GPT-5.4: 56.4 (#22), Qwen3.6 Max Preview: 54.3 (#46)
| Benchmark | GPT-5.4 | Qwen3.6 Max Preview |
|---|---|---|
| LMArena Non-English | 1465 | 1436 |
| LMArena Chinese | 1520 | 1487 |
| LMArena French | 1493 | 1449 |
| LMArena Russian | 1480 | 1444 |
| LMArena Spanish | 1453 | 1452 |
| LMArena German | 1472 | — |
| LMArena Japanese | 1484 | — |
| LMArena Korean | 1449 | — |
Instruction Following GPT-5.4 leads
GPT-5.4: 77.3 (#25), Qwen3.6 Max Preview: 75.8 (#51)
| Benchmark | GPT-5.4 | Qwen3.6 Max Preview |
|---|---|---|
| LMArena Instruction Following | 1469 | 1437 |
Long Context GPT-5.4 leads
GPT-5.4: 50.3 (#8), Qwen3.6 Max Preview: 44.6 (#59)
| Benchmark | GPT-5.4 | Qwen3.6 Max Preview |
|---|---|---|
| LMArena Longer Query | 1474 | 1457 |
| CL-bench | 27.9% | — |
| CL-bench Life | 21.7% | — |
Writing & Preference GPT-5.4 leads
GPT-5.4: 72.1 (#18), Qwen3.6 Max Preview: 64.0 (#58)
| Benchmark | GPT-5.4 | Qwen3.6 Max Preview |
|---|---|---|
| LMArena Text | 1469 | 1447 |
| LMArena Creative Writing | 1439 | 1435 |
| LMArena Multi-Turn | 1482 | 1456 |
| EQ-Bench Creative Writing | 1840 | — |
| EQ-Bench 4 | 1272 | — |
Frequently asked questions
Is GPT-5.4 better than Qwen3.6 Max Preview?
GPT-5.4 is the stronger model overall, scoring 59.5 to 51.1 on the Noometry Index. Qwen3.6 Max Preview costs 1.9× less per token, which makes it the better buy when GPT-5.4's lead doesn't matter for your workload.
Which is cheaper, GPT-5.4 or Qwen3.6 Max Preview?
Qwen3.6 Max Preview is cheaper. It lists at $1.30 per million input tokens and $7.80 per million output tokens; GPT-5.4 lists at $2.50 and $15.
Is GPT-5.4 or Qwen3.6 Max Preview better for coding?
GPT-5.4 scores higher on coding benchmarks: 52.8 versus 48.6 in the Noometry coding category.
Which has the bigger context window?
GPT-5.4 does, with 1.05M tokens against 262K.
How many benchmarks do GPT-5.4 and Qwen3.6 Max Preview share?
28 benchmarks have published results for both models. GPT-5.4 has 68 scored results on Noometry and Qwen3.6 Max Preview has 29.