Across 30 shared benchmarks, Kimi K2 (Reasoning) scores higher on 12 and Qwen3.5 122B A10B on 16, with 2 level. The widest gap is critpt, where Kimi K2 (Reasoning) scores 2.6 against 0.9. Tracked API pricing per million tokens: Kimi K2 (Reasoning) $0.60 in / $2.50 out, Qwen3.5 122B A10B $0.40 in / $3.20 out.
| Benchmark | Kimi K2 (Reasoning) | Qwen3.5 122B A10B |
|---|---|---|
| AA Agentic Index | 47.9 | 21.3 |
| AA Intelligence | 33.5 | 32.8 |
| AA-LCR | 70.3 | 70.3 |
| AA-Omniscience | -21.4 | -41.5 |
| Artificial Analysis Coding Index | 34.8 | 45.7 |
| browsecomp | 60.2 | 63.8 |
| browsecomp_zh | 62.3 | 69.9 |
| critpt | 2.6 | 0.9 |
| gdpval | 24.5 | 24.3 |
| GPQA Diamond | 84.5 | 86.6 |
| HLE | 23.9 | 47.5 |
| HMMT 2025 | 89.4 | 90.3 |
| IFBench | 68.1 | 76.1 |
| LiveCodeBench v6 | 83.1 | 78.9 |
| longbench_v2 | 45.1 | 60.2 |
| mmlu_redux | 94.4 | 94 |
| MMLU-Pro | 84.6 | 86.7 |
| multichallenge | 66.4 | 61.5 |
| OmniScience Accuracy | 30.9 | 24.4 |
| OmniScience Non-Hallucination | 25.8 | 12.9 |
| scicode | 44.8 | 42 |
| Seal-0 | 56.3 | 44.1 |
| SWE-bench Verified | 71.3 | 72 |
| Terminal-Bench 2.0 | 35.7 | 49.4 |
| Terminal-Bench Hard | 31.1 | 31.1 |
| vectara_answer_rate | 98.6 | 99.8 |
| vectara_avg_summary_length | 59.2 | 86.4 |
| vectara_factual_consistency | 82.1 | 88.8 |
| vectara_hallucination_rate ↓ | 17.9 | 11.2 |
| τ²-Bench Telecom (AA run) | 93 | 93.6 |
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.