xbench-DeepSearch — Rollup spanning multiple vintages per its own raws list; not pinned to one snapshot
| # | Model | Vendor | Best score | Runs | Last seen |
|---|---|---|---|---|---|
| 1 | GPT-5 | OpenAI | 77.8 | 1 | 2026-06-06 |
| 2 | Kimi K2.5 | Moonshot | 76.7 | 3 | 2026-05-30 |
| 3 | Kimi K2 (Reasoning) | Moonshot | 76 | 4 | 2026-06-06 |
| 4 | MiniMax M2 | MiniMax | 72 | 1 | 2026-06-06 |
| 5 | DeepSeek-V3.2 | DeepSeek | 71 | 4 | 2026-06-06 |
| 6 | GLM-4.6 | Z.ai | 70 | 1 | 2026-06-06 |
| 7 | Claude Sonnet 4.5 | Anthropic | 66 | 1 | 2026-06-06 |
| 8 | Claude Sonnet 4 | Anthropic | 64.6 | 1 | 2026-06-06 |
| 9 | Step 3.5 Flash | StepFun | 56.3 | 3 | 2026-05-30 |
| 10 | Gemini 2.5 Pro | 56 | 1 | 2026-06-06 | |
| 11 | GLM-4.7 | Z.ai | 52.3 | 3 | 2026-05-30 |
| 12 | MiMo V2 Flash | Xiaomi | 44 | 3 | 2026-05-30 |
| 13 | MiniMax M2.1 | MiniMax | 43 | 3 | 2026-05-30 |
Best tracked score per model (default configuration; source-attributed and verification-tiered). Open a model for its full benchmark surface, provenance and pricing.