BlogLLM Benchmark Scores as Published, Side by Side
Kimi K2 Base: Benchmark Scores as Published
Every result the catalog holds for Kimi K2 Base: 20 tests in 5 areas, from 1 page. Each was printed by the company named beside it. Each number is as printed; nothing is ranked or rescored.
Developer Moonshot AIReleased 2025-07-11Context window 128KInput TextOutput TextWeights Open (Modified MIT License)Data version 2026103001, as of 2026-10-30
Agentic Coding
| Test | Value | Printed by | Setup as printed |
|---|---|---|---|
| SWE-bench (agentless, split not stated) | 28.2% | Xiaomi, Xiaomi MiMo-V2.5-Pro model cardchecked 2026-10-15 | Harness: 3-shot |
Science and Math
| Test | Value | Printed by | Setup as printed |
|---|---|---|---|
| GPQA Diamond | 48.1% | Xiaomi, Xiaomi MiMo-V2.5-Pro model cardchecked 2026-10-15 | Harness: 5-shot |
| MATH | 70.2% | Xiaomi, Xiaomi MiMo-V2.5-Pro model cardchecked 2026-10-15 | Harness: 4-shot |
| GSM8K | 92.1% | Xiaomi, Xiaomi MiMo-V2.5-Pro model cardchecked 2026-10-15 | Harness: 8-shot |
| AIME 2024 and 2025 | 31.6% | Xiaomi, Xiaomi MiMo-V2.5-Pro model cardchecked 2026-10-15 | Harness: 2-shot |
General Knowledge and Reasoning
| Test | Value | Printed by | Setup as printed |
|---|---|---|---|
| MMLU-Pro | 69.2% | Xiaomi, Xiaomi MiMo-V2.5-Pro model cardchecked 2026-10-15 | Harness: 5-shot |
| MMLU-Redux | 90.2% | Xiaomi, Xiaomi MiMo-V2.5-Pro model cardchecked 2026-10-15 | Harness: 5-shot |
| C-Eval | 92.5% | Xiaomi, Xiaomi MiMo-V2.5-Pro model cardchecked 2026-10-15 | Harness: 5-shot |
| MMLU | 87.8% | Xiaomi, Xiaomi MiMo-V2.5-Pro model cardchecked 2026-10-15 | Harness: 5-shot |
| DROP (F1) | 83.6% | Xiaomi, Xiaomi MiMo-V2.5-Pro model cardchecked 2026-10-15 | Harness: 3-shot |
| BBH | 88.7% | Xiaomi, Xiaomi MiMo-V2.5-Pro model cardchecked 2026-10-15 | Harness: 3-shot |
| CMMLU | 90.9% | Xiaomi, Xiaomi MiMo-V2.5-Pro model cardchecked 2026-10-15 | Harness: 5-shot |
| HellaSwag | 94.6% | Xiaomi, Xiaomi MiMo-V2.5-Pro model cardchecked 2026-10-15 | Harness: 10-shot |
| TriviaQA | 85.1% | Xiaomi, Xiaomi MiMo-V2.5-Pro model cardchecked 2026-10-15 | Harness: 5-shot |
| WinoGrande | 85.3% | Xiaomi, Xiaomi MiMo-V2.5-Pro model cardchecked 2026-10-15 | Harness: 5-shot |
| ARC-Challenge | 96.2% | Xiaomi, Xiaomi MiMo-V2.5-Pro model cardchecked 2026-10-15 | Harness: 25-shot |
Code Generation
| Test | Value | Printed by | Setup as printed |
|---|---|---|---|
| LiveCodeBench v6 | 26.3% | Xiaomi, Xiaomi MiMo-V2.5-Pro model cardchecked 2026-10-15 | Harness: 1-shot |
| HumanEval+ | 84.8% | Xiaomi, Xiaomi MiMo-V2.5-Pro model cardchecked 2026-10-15 | Harness: 1-shot |
| MBPP+ | 73.8% | Xiaomi, Xiaomi MiMo-V2.5-Pro model cardchecked 2026-10-15 | Harness: 3-shot |
Multilingual
| Test | Value | Printed by | Setup as printed |
|---|---|---|---|
| Global-MMLU | 80.7% | Xiaomi, Xiaomi MiMo-V2.5-Pro model cardchecked 2026-10-15 | Harness: 5-shot |
More from Moonshot AI
Every number here is read from the same catalog file as the benchmark table, data version 2026103001. Where one page printed the model under several setups, each value is listed. The catalog is public in armanobosyan/sugra-llm-benchmarks; a wrong number can be reported from the table, with Report an error under the value, or as an issue in that repository.