Across 33 shared benchmarks, Gemini 3 Flash Preview scores higher on 23 and Qwen3.5 122B A10B on 9, with 1 level. The widest gap is critpt, where Gemini 3 Flash Preview scores 8.6 against 0.9. Tracked API pricing per million tokens: Gemini 3 Flash Preview $0.50 in / $3.00 out, Qwen3.5 122B A10B $0.40 in / $3.20 out.
| Benchmark | Gemini 3 Flash Preview | Qwen3.5 122B A10B |
|---|---|---|
| AA Agentic Index | 49.7 | 21.3 |
| AA Intelligence | 38.7 | 32.8 |
| AA-LCR | 73 | 70.3 |
| AA-Omniscience | 10.1 | -41.5 |
| arena_vision | 1265 | 1246 |
| Artificial Analysis Coding Index | 42.6 | 45.7 |
| coding_arena_elo | 1387 | 1358 |
| critpt | 8.6 | 0.9 |
| gdpval | 35.2 | 24.3 |
| GPQA Diamond | 90.4 | 86.6 |
| HLE | 43.5 | 47.5 |
| IFBench | 78 | 76.1 |
| mmmlu | 91.8 | 86.7 |
| MMMU-Pro | 81.2 | 76.9 |
| OmniDocBench 1.5 | 12.1 | 89.8 |
| OmniScience Accuracy | 53.4 | 24.4 |
| OmniScience Non-Hallucination | 7.6 | 12.9 |
| OSWorld-Verified | 65.1 | 58 |
| scicode | 50.6 | 42 |
| screenspot_pro_no_tools | 69.1 | 70.4 |
| SWE-bench Verified | 78 | 72 |
| t2-bench | 90.2 | 79.5 |
| TauBench V3 - Banking | 20.8 | 15.3 |
| Terminal-Bench 2.0 | 47.6 | 49.4 |
| Terminal-Bench 2.1 | 58 | 47.6 |
| Terminal-Bench Hard | 38.6 | 31.1 |
| vectara_answer_rate | 99.8 | 99.8 |
| vectara_avg_summary_length | 90.2 | 86.4 |
| vectara_factual_consistency | 86.5 | 88.8 |
| vectara_hallucination_rate ↓ | 13.5 | 11.2 |
| VideoMMMU | 86.9 | 82 |
| τ²-Bench Telecom (AA run) | 80.4 | 93.6 |
| τ³-Bench | 17.5 | 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.