Across 39 shared benchmarks, GPT-5 scores higher on 27 and Qwen3.5 122B A10B on 12. The widest gap is critpt, where GPT-5 scores 5.7 against 0.9. Qwen3.5 122B A10B is the cheaper of the two on tracked API pricing ($0.40 against $1.25 per million input tokens).
| Benchmark | GPT-5 | Qwen3.5 122B A10B |
|---|---|---|
| AA Agentic Index | 49.7 | 21.3 |
| AA Intelligence | 35.3 | 32.8 |
| AA-LCR | 76.3 | 70.3 |
| AA-Omniscience | -8.7 | -41.5 |
| arena_vision | 1212 | 1246 |
| Artificial Analysis Coding Index | 38.9 | 45.7 |
| browsecomp | 54.9 | 63.8 |
| browsecomp_zh | 65 | 69.9 |
| coding_arena_elo | 1419 | 1358 |
| critpt | 5.7 | 0.9 |
| ERQA | 65.7 | 62 |
| gdpval | 32.5 | 24.3 |
| GPQA Diamond | 87.3 | 86.6 |
| GSM8K | -1.8 | 94.5 |
| HLE | 28.5 | 47.5 |
| HMMT 2025 | 93.3 | 90.3 |
| IFBench | 73.1 | 76.1 |
| LiveCodeBench v6 | 87 | 78.9 |
| mmlu_redux | 95.3 | 94 |
| MMLU-Pro | 87.5 | 86.7 |
| MMMU | 84.2 | 83.9 |
| MMMU-Pro | 78.4 | 76.9 |
| multichallenge | 71.1 | 61.5 |
| OmniScience Accuracy | 40.3 | 24.4 |
| OmniScience Non-Hallucination | 21 | 12.9 |
| scicode | 43 | 42 |
| Seal-0 | 51.4 | 44.1 |
| SWE-bench Verified | 74.9 | 72 |
| TauBench V3 - Banking | 22.1 | 15.3 |
| Terminal-Bench 2.0 | 35.2 | 49.4 |
| Terminal-Bench 2.1 | 35.2 | 47.6 |
| Terminal-Bench Hard | 37.9 | 31.1 |
| vectara_answer_rate | 99.9 | 99.8 |
| vectara_avg_summary_length | 162.7 | 86.4 |
| vectara_factual_consistency | 85.3 | 88.8 |
| vectara_hallucination_rate ↓ | 14.7 | 11.2 |
| VideoMMMU | 84.6 | 82 |
| τ²-Bench Telecom (AA run) | 86.5 | 93.6 |
| τ³-Bench | 19.6 | 13.6 |
Best tracked score per model per benchmark (default configuration; source-attributed). ↓ marks lower-is-better metrics. Open either model for its full surface, provenance and pricing. Quoted rates are the price-setter row we currently track for each model — its direct or vendor-official listing where one exists (direct, direct), otherwise the lowest tracked offer.