BlogLLM Benchmark Scores as Published, Side by Side
Kimi K2 Instruct: Benchmark Scores as Published
Every result the catalog holds for Kimi K2 Instruct: 31 tests in 6 areas, from 1 page. All were printed by Moonshot AI itself. 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 Verified | 65.8% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | Harness: agentic, single attempt |
| SWE-bench Multilingual | 47.3% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | Harness: agentic, single attempt |
| Terminal-Bench (version not stated) | 30% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | Harness: in-house framework |
| 25% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | Harness: Terminus | |
| SWE-bench Verified (agentless, single patch without test) | 51.8% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | Harness: agentless |
| SWE-bench Verified (multiple attempts) | 71.6% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | Harness: agentic, multiple attempts |
Science and Math
| Test | Value | Printed by | Setup as printed |
|---|---|---|---|
| GPQA Diamond | 75.1% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | - |
| AIME 2025 | 49.5% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | - |
| HMMT February 2025 | 38.8% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | - |
| AIME 2024 | 69.6% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | - |
| MATH-500 | 97.4% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | - |
| CNMO 2024 | 74.3% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | - |
| PolyMath (en) | 65.1% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | - |
General Knowledge and Reasoning
| Test | Value | Printed by | Setup as printed |
|---|---|---|---|
| Humanity's Last Exam (no tools) | 4.7% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | - |
| MMLU-Pro | 81.1% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | - |
| MMLU-Redux | 92.7% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | - |
| SuperGPQA | 57.2% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | - |
| MMLU | 89.5% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | - |
| SimpleQA | 31% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | - |
| AutoLogi | 89.5% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | - |
| LiveBench (2024-11-25) | 76.4% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | - |
| ZebraLogic | 89% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | - |
Code Generation
| Test | Value | Printed by | Setup as printed |
|---|---|---|---|
| LiveCodeBench v6 | 53.7% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | - |
| Aider polyglot | 60% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | - |
| MultiPL-E | 85.7% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | - |
| OJBench (language not stated) | 27.1% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | - |
Tool Use and Workflows
| Test | Value | Printed by | Setup as printed |
|---|---|---|---|
| τ²-bench Telecom | 65.8% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | - |
| τ²-bench Retail | 70.6% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | - |
| τ²-bench Airline | 56.5% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | - |
| AceBench | 76.5% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | - |
Instruction Following
| Test | Value | Printed by | Setup as printed |
|---|---|---|---|
| MultiChallenge | 54.1% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | - |
| IFEval (prompt-level strict) | 89.8% | Moonshot AI, Kimi K2 release pagechecked 2026-10-15 | - |
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.