Model comparison
GPT-5 Mini vs Qwen3.8 Max
Qwen3.8 Max is the stronger model overall, scoring 56.8 to 41.8 on the Noometry Index. GPT-5 Mini costs 4.4× less per token, which makes it the better buy when Qwen3.8 Max's lead doesn't matter for your workload.
Last verified . 31 shared benchmarks.
Alibaba (Qwen)Summary
- They share 31 benchmarks with published results for both. GPT-5 Mini scores higher in 0 categories and Qwen3.8 Max in 10 categories; 10 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where Qwen3.8 Max leads 54.4 to 23.9.
- The biggest single-benchmark swing is ProofBench: 9% for GPT-5 Mini and 58% for Qwen3.8 Max.
- GPT-5 Mini is cheaper at $0.25 / $2 per million input/output tokens, against $2 / $6 for Qwen3.8 Max.
- Qwen3.8 Max accepts more context: 1M tokens versus 400K.
Side by side
| GPT-5 Mini | Qwen3.8 Max | |
|---|---|---|
| Provider | OpenAI | Alibaba (Qwen) |
| Noometry Index | 41.8 | 56.8 |
| Released | 2025-08-07 | 2026-08-02 |
| Weights | Proprietary | Proprietary |
| Context window | 400K | 1M |
| Max output | 128K | 131K |
| Input $ / M tokens | $0.25 | $2 |
| Output $ / M tokens | $2 | $6 |
| Results tracked | 60 | 39 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Qwen3.8 Max leads
GPT-5 Mini: 40.1 (#146), Qwen3.8 Max: 53.5 (#29)
| Benchmark | GPT-5 Mini | Qwen3.8 Max |
|---|---|---|
| SciCode | 39.2% | 53.2% |
| LMArena Coding | 1406 | 1502 |
| SWE-bench Verified | 64.7% | — |
| DeepSWE | — | 57.5% |
| SWE-bench Verified (bash only) | 59.8% | — |
| LMArena WebDev | — | 1674 |
| SWE-bench Multilingual | 39.7% | — |
| FrontierSWE | — | 17.8% |
| WeirdML | 52.7% | — |
| ALE-Bench | 799.77 | — |
| AlgoTune | 1.38 | — |
Agentic & Tool Use Qwen3.8 Max leads
GPT-5 Mini: 31.1 (#70), Qwen3.8 Max: 45.4 (#14)
| Benchmark | GPT-5 Mini | Qwen3.8 Max |
|---|---|---|
| Terminal-Bench | 34.8% | — |
| APEX-Agents | — | 63.3% |
| Berkeley Function Calling Leaderboard | 55.5% | — |
| τ²-bench Banking | — | 55.1% |
| GDP.pdf | — | 23.2% |
| Vending-Bench 2 | -31.18 | — |
Reasoning Qwen3.8 Max leads
GPT-5 Mini: 23.9 (#168), Qwen3.8 Max: 54.4 (#26)
| Benchmark | GPT-5 Mini | Qwen3.8 Max |
|---|---|---|
| CritPt | 0% | 20% |
| Chess Puzzles | 30% | 40% |
| LMArena Hard Prompts | 1380 | 1496 |
| Mystery Game Puzzles | 10% | 38% |
| DTBench | 80.5% | 92% |
| LMCA | 34.2% | 46.2% |
| Epoch Capabilities Index | 145.52 | 156.41 |
| ARC-AGI-2 | 4.4% | — |
| Kagi LLM Benchmark | 70.3% | — |
| NYT Connections (extended) | — | 88.3% |
| ARC-AGI-1 | 54.3% | — |
| EnigmaEval | 8.2% | — |
| ForecastBench | 61 | — |
Math Qwen3.8 Max leads
GPT-5 Mini: 46.7 (#69), Qwen3.8 Max: 73.2 (#20)
| Benchmark | GPT-5 Mini | Qwen3.8 Max |
|---|---|---|
| FrontierMath (Tiers 1-3) | 46.7% | 74.7% |
| FrontierMath Tier 4 | 12.2% | 46.3% |
| OTIS Mock AIME 2024-2025 | 86.7% | 100% |
| ProofBench | 9% | 58% |
| LMArena Math | 1378 | 1499 |
| Omni-MATH | 72.2% | — |
| MATH Level 5 | 97.8% | — |
| FrontierMath (Feb 2025 set) | 27.2% | — |
| FrontierMath Tier 4 (v1) | 6.3% | — |
Knowledge Qwen3.8 Max leads
GPT-5 Mini: 45.6 (#86), Qwen3.8 Max: 61.7 (#27)
| Benchmark | GPT-5 Mini | Qwen3.8 Max |
|---|---|---|
| GPQA Diamond | 75% | 92.7% |
| SimpleQA Verified | 21.6% | 47.3% |
| LMArena Expert | 1379 | 1507 |
| Humanity's Last Exam | 19.4% | — |
| MMLU-Pro | 83.5% | — |
| Confabulations | 13.3% | — |
| Vectara Hallucination Rate | 12.9% | — |
| GPQA (HELM) | 75.6% | — |
Multimodal Qwen3.8 Max leads
GPT-5 Mini: 35.6 (#85), Qwen3.8 Max: 37.2 (#75)
| Benchmark | GPT-5 Mini | Qwen3.8 Max |
|---|---|---|
| LMArena Vision | 1202 | 1314 |
| VPCT | 40.2% | — |
| Furniture Assembly | — | 20% |
Multilingual Qwen3.8 Max leads
GPT-5 Mini: 48.9 (#137), Qwen3.8 Max: 56.7 (#18)
| Benchmark | GPT-5 Mini | Qwen3.8 Max |
|---|---|---|
| LMArena Non-English | 1363 | 1472 |
| LMArena Chinese | 1385 | 1538 |
| LMArena French | 1386 | 1503 |
| LMArena German | 1366 | 1483 |
| LMArena Japanese | 1341 | 1467 |
| LMArena Korean | 1308 | 1461 |
| LMArena Russian | 1362 | 1481 |
| LMArena Spanish | 1355 | 1492 |
Instruction Following Qwen3.8 Max leads
GPT-5 Mini: 76.2 (#46), Qwen3.8 Max: 77.6 (#17)
| Benchmark | GPT-5 Mini | Qwen3.8 Max |
|---|---|---|
| LMArena Instruction Following | 1357 | 1479 |
| IFEval | 92.7% | — |
Long Context Qwen3.8 Max leads
GPT-5 Mini: 41.9 (#132), Qwen3.8 Max: 45.6 (#31)
| Benchmark | GPT-5 Mini | Qwen3.8 Max |
|---|---|---|
| LMArena Longer Query | 1355 | 1489 |
| Fiction.LiveBench | 69.4% | — |
Writing & Preference Qwen3.8 Max leads
GPT-5 Mini: 55.2 (#148), Qwen3.8 Max: 67.1 (#30)
| Benchmark | GPT-5 Mini | Qwen3.8 Max |
|---|---|---|
| LMArena Text | 1373 | 1483 |
| LMArena Creative Writing | 1325 | 1479 |
| LMArena Multi-Turn | 1363 | 1489 |
| Short-Story Creative Writing | 83.1% | — |
| EQ-Bench Creative Writing | 1313 | — |
| WildBench | 85.5% | — |
Frequently asked questions
Is GPT-5 Mini better than Qwen3.8 Max?
Qwen3.8 Max is the stronger model overall, scoring 56.8 to 41.8 on the Noometry Index. GPT-5 Mini costs 4.4× less per token, which makes it the better buy when Qwen3.8 Max's lead doesn't matter for your workload.
Which is cheaper, GPT-5 Mini or Qwen3.8 Max?
GPT-5 Mini is cheaper. It lists at $0.25 per million input tokens and $2 per million output tokens; Qwen3.8 Max lists at $2 and $6.
Is GPT-5 Mini or Qwen3.8 Max better for coding?
Qwen3.8 Max scores higher on coding benchmarks: 53.5 versus 40.1 in the Noometry coding category.
Which has the bigger context window?
Qwen3.8 Max does, with 1M tokens against 400K.
How many benchmarks do GPT-5 Mini and Qwen3.8 Max share?
31 benchmarks have published results for both models. GPT-5 Mini has 60 scored results on Noometry and Qwen3.8 Max has 39.