Model comparison
GPT-5.4 mini vs Qwen3.5 397B-A17B
Qwen3.5 397B-A17B is the stronger model overall, scoring 46.0 to 45.0 on the Noometry Index.
Last verified . 31 shared benchmarks.
Alibaba (Qwen)Summary
- They share 31 benchmarks with published results for both. GPT-5.4 mini scores higher in 2 categories and Qwen3.5 397B-A17B in 8 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where Qwen3.5 397B-A17B leads 34.5 to 30.4.
- The biggest single-benchmark swing is Kagi LLM Benchmark: 37.9% for GPT-5.4 mini and 73.7% for Qwen3.5 397B-A17B.
- Qwen3.5 397B-A17B is cheaper at $0.60 / $3.60 per million input/output tokens, against $0.75 / $4.50 for GPT-5.4 mini.
- GPT-5.4 mini accepts more context: 400K tokens versus 262K.
- Qwen3.5 397B-A17B has downloadable open weights; the other is API-only.
Side by side
| GPT-5.4 mini | Qwen3.5 397B-A17B | |
|---|---|---|
| Provider | OpenAI | Alibaba (Qwen) |
| Noometry Index | 45.0 | 46.0 |
| Released | 2026-03-17 | 2026-02-01 |
| Weights | Proprietary | Open |
| Context window | 400K | 262K |
| Max output | 128K | 66K |
| Input $ / M tokens | $0.75 | $0.60 |
| Output $ / M tokens | $4.50 | $3.60 |
| Results tracked | 46 | 36 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-5.4 mini leads
GPT-5.4 mini: 45.2 (#72), Qwen3.5 397B-A17B: 42.0 (#114)
| Benchmark | GPT-5.4 mini | Qwen3.5 397B-A17B |
|---|---|---|
| LMArena WebDev | 1397 | 1400 |
| LMArena Coding | 1438 | 1465 |
| FrontierCode | 27% | — |
| SciCode | 49.9% | — |
| WeirdML | 60.3% | — |
| ALE-Bench | 1,189 | — |
Agentic & Tool Use Qwen3.5 397B-A17B leads
GPT-5.4 mini: 29.9 (#81), Qwen3.5 397B-A17B: 33.3 (#53)
| Benchmark | GPT-5.4 mini | Qwen3.5 397B-A17B |
|---|---|---|
| APEX-Agents | — | 24.9% |
| τ²-bench Airline | — | 81.5% |
| τ²-bench Banking | — | 9.8% |
| τ²-bench Retail | — | 84.4% |
| τ²-bench Telecom | — | 97.8% |
| DeepResearch Bench | 36.3% | — |
Reasoning Qwen3.5 397B-A17B leads
GPT-5.4 mini: 30.4 (#85), Qwen3.5 397B-A17B: 34.5 (#70)
| Benchmark | GPT-5.4 mini | Qwen3.5 397B-A17B |
|---|---|---|
| Kagi LLM Benchmark | 37.9% | 73.7% |
| NYT Connections (extended) | 61.8% | 58.9% |
| Chess Puzzles | 24% | 13% |
| Thematic Generalization | 61.7% | 65.1% |
| LMArena Hard Prompts | 1424 | 1448 |
| Mystery Game Puzzles | 11% | 18% |
| DTBench | 80% | 87.5% |
| LMCA | 40.8% | 37.9% |
| Epoch Capabilities Index | 148.84 | 146.65 |
| ARC-AGI-2 | 18.9% | — |
| ARC-AGI-1 | 63.7% | — |
| CritPt | 10% | — |
| ForecastBench | 57 | — |
Math Too close to call
GPT-5.4 mini: 45.5 (#75), Qwen3.5 397B-A17B: 46.1 (#73)
| Benchmark | GPT-5.4 mini | Qwen3.5 397B-A17B |
|---|---|---|
| FrontierMath (Tiers 1-3) | 51.2% | 31.2% |
| OTIS Mock AIME 2024-2025 | 88.9% | 88.9% |
| LMArena Math | 1419 | 1454 |
| FrontierMath Tier 4 | 9.8% | — |
| ProofBench | 21% | — |
| FrontierMath (Feb 2025 set) | 28.3% | — |
| FrontierMath Tier 4 (v1) | 2.1% | — |
Knowledge Qwen3.5 397B-A17B leads
GPT-5.4 mini: 51.5 (#67), Qwen3.5 397B-A17B: 53.3 (#58)
| Benchmark | GPT-5.4 mini | Qwen3.5 397B-A17B |
|---|---|---|
| GPQA Diamond | 86.9% | 86.4% |
| LMArena Expert | 1435 | 1462 |
| SimpleQA Verified | 29.4% | — |
| Vectara Hallucination Rate | 5.5% | — |
Multimodal Qwen3.5 397B-A17B leads
GPT-5.4 mini: 39.7 (#56), Qwen3.5 397B-A17B: 40.7 (#44)
| Benchmark | GPT-5.4 mini | Qwen3.5 397B-A17B |
|---|---|---|
| LMArena Vision | 1245 | 1263 |
Multilingual Qwen3.5 397B-A17B leads
GPT-5.4 mini: 51.9 (#96), Qwen3.5 397B-A17B: 53.7 (#59)
| Benchmark | GPT-5.4 mini | Qwen3.5 397B-A17B |
|---|---|---|
| LMArena Non-English | 1405 | 1430 |
| LMArena Chinese | 1446 | 1500 |
| LMArena French | 1440 | 1461 |
| LMArena German | 1409 | 1447 |
| LMArena Japanese | 1374 | 1426 |
| LMArena Korean | 1368 | 1384 |
| LMArena Russian | 1417 | 1429 |
| LMArena Spanish | 1405 | 1441 |
Instruction Following Too close to call
GPT-5.4 mini: 74.1 (#102), Qwen3.5 397B-A17B: 75.0 (#77)
| Benchmark | GPT-5.4 mini | Qwen3.5 397B-A17B |
|---|---|---|
| LMArena Instruction Following | 1405 | 1424 |
Long Context Qwen3.5 397B-A17B leads
GPT-5.4 mini: 43.0 (#112), Qwen3.5 397B-A17B: 44.1 (#74)
| Benchmark | GPT-5.4 mini | Qwen3.5 397B-A17B |
|---|---|---|
| LMArena Longer Query | 1407 | 1442 |
Writing & Preference GPT-5.4 mini leads
GPT-5.4 mini: 64.0 (#58), Qwen3.5 397B-A17B: 62.3 (#79)
| Benchmark | GPT-5.4 mini | Qwen3.5 397B-A17B |
|---|---|---|
| LMArena Text | 1412 | 1438 |
| LMArena Creative Writing | 1370 | 1401 |
| EQ-Bench Creative Writing | 1665 | 1478 |
| LMArena Multi-Turn | 1429 | 1446 |
Frequently asked questions
Is GPT-5.4 mini better than Qwen3.5 397B-A17B?
Qwen3.5 397B-A17B is the stronger model overall, scoring 46.0 to 45.0 on the Noometry Index.
Which is cheaper, GPT-5.4 mini or Qwen3.5 397B-A17B?
Qwen3.5 397B-A17B is cheaper. It lists at $0.60 per million input tokens and $3.60 per million output tokens; GPT-5.4 mini lists at $0.75 and $4.50.
Is GPT-5.4 mini or Qwen3.5 397B-A17B better for coding?
GPT-5.4 mini scores higher on coding benchmarks: 45.2 versus 42.0 in the Noometry coding category.
Which has the bigger context window?
GPT-5.4 mini does, with 400K tokens against 262K.
How many benchmarks do GPT-5.4 mini and Qwen3.5 397B-A17B share?
31 benchmarks have published results for both models. GPT-5.4 mini has 46 scored results on Noometry and Qwen3.5 397B-A17B has 36.