BlogLLM Benchmark Scores as Published, Side by Side
Kimi K2.5: Benchmark Scores as Published
Every result the catalog holds for Kimi K2.5: 119 tests in 16 areas, from 7 pages. 55 printed by Moonshot AI itself, the rest by the company named beside each number. Each number is as printed; nothing is ranked or rescored.
Developer Moonshot AIReleased 2026-01-27Context window 256KInput Image, videoOutput TextWeights Open (Modified MIT License)Data version 2026103001, as of 2026-10-30
Knowledge Work
| Test | Value | Printed by | Setup as printed |
|---|---|---|---|
| GDPval-AA v2 | 1009 | Thinking Machines, Thinking Machines Inkling model cardchecked 2026-10-15 | - |
| APEX-Agents | 11.5% | Moonshot AI, Kimi K2.6 model cardchecked 2026-10-15 | - |
| Vending-Bench 2 (final balance in USD) | $1198.46 | Z.ai, GLM-5 model cardchecked 2026-10-15; printed on 1 more page | - |
Agentic Coding
| Test | Value | Printed by | Setup as printed |
|---|---|---|---|
| SWE-bench Verified | 76.8% | Moonshot AI, Kimi K2.5 model cardchecked 2026-10-15; printed on 5 more pages | Harness: in-house SWE-agent variant |
| SWE-Bench Pro | 50.7% | Moonshot AI, Kimi K2.5 model cardchecked 2026-10-15; printed on 3 more pages | Harness: in-house SWE-agent variant |
| Terminal-Bench 2.1 | 51.3% | Thinking Machines, Thinking Machines Inkling model cardchecked 2026-10-15 | Harness: best harness per model, as printed |
| SWE-bench Multilingual | 73% | Moonshot AI, Kimi K2.5 model cardchecked 2026-10-15; printed on 3 more pages | Harness: in-house SWE-agent variant |
| Terminal-Bench 2.0 | 50.8% | Moonshot AI, Kimi K2.5 model cardchecked 2026-10-15; printed on 4 more pages | Harness: Terminus-2 |
| NL2Repo-Bench | 32 | Z.ai, GLM-5.1 model cardchecked 2026-10-15 | - |
ML Engineering
| Test | Value | Printed by | Setup as printed |
|---|---|---|---|
| PaperBench | 63.5 | Moonshot AI, Kimi K2.5 model cardchecked 2026-10-15 | - |
Science and Math
| Test | Value | Printed by | Setup as printed |
|---|---|---|---|
| GPQA Diamond | 87.6% | Moonshot AI, Kimi K2.5 model cardchecked 2026-10-15; printed on 5 more pages | - |
| AIME 2025 | 96.1% | Moonshot AI, Kimi K2.5 model cardchecked 2026-10-15; printed on 1 more page | - |
| SciCode | 48.7% | Moonshot AI, Kimi K2.5 model cardchecked 2026-10-15; printed on 1 more page | - |
| HMMT February 2025 | 95.4% | Moonshot AI, Kimi K2.5 model cardchecked 2026-10-15; printed on 1 more page | - |
| IMO-AnswerBench | 81.8% | Moonshot AI, Kimi K2.5 model cardchecked 2026-10-15; printed on 4 more pages | - |
| AIME 2026 | 95.8% | Moonshot AI, Kimi K2.6 model cardchecked 2026-10-15; printed on 4 more pages | - |
| HMMT November 2025 | 91.1% | Alibaba, Qwen3.5-397B-A17B model cardchecked 2026-10-15; printed on 2 more pages | - |
| HMMT February 2026 | 87.1% | Moonshot AI, Kimi K2.6 model cardchecked 2026-10-15; printed on 1 more page | - |
| PolyMATH | 43.1% | Alibaba, Qwen3.5-397B-A17B model cardchecked 2026-10-15 | - |
| BeyondAIME | 60.3% | Mistral AI, Mistral Medium 3.5 model cardchecked 2026-10-15 | Effort: max |
Long Context
| Test | Value | Printed by | Setup as printed |
|---|---|---|---|
| AA-LCR | 70% | Alibaba, Qwen3.5-397B-A17B model cardchecked 2026-10-15 | - |
| LongBench v2 | 61% | Moonshot AI, Kimi K2.5 model cardchecked 2026-10-15; printed on 1 more page | Context: 128000 |
Computer Use
| Test | Value | Printed by | Setup as printed |
|---|---|---|---|
| OSWorld-Verified | 63.3% | Moonshot AI, Kimi K2.6 model cardchecked 2026-10-15; printed on 1 more page | - |
Multimodal Understanding
Cybersecurity
| Test | Value | Printed by | Setup as printed |
|---|---|---|---|
| CyberGym | 41.3% | Moonshot AI, Kimi K2.5 model cardchecked 2026-10-15; printed on 2 more pages | - |
| SecCodeBench | 61.3% | Alibaba, Qwen3.5-397B-A17B model cardchecked 2026-10-15 | - |
General Knowledge and Reasoning
| Test | Value | Printed by | Setup as printed |
|---|---|---|---|
| Humanity's Last Exam (no tools) | 30.1% | Moonshot AI, Kimi K2.5 model cardchecked 2026-10-15; printed on 5 more pages | Tools: none |
| MMLU-Pro | 87.1% | Moonshot AI, Kimi K2.5 model cardchecked 2026-10-15; printed on 1 more page | - |
| Humanity's Last Exam (with tools) | 50.2% | Moonshot AI, Kimi K2.5 model cardchecked 2026-10-15; printed on 5 more pages | Tools: search, code interpreter, browsing |
| MMLU-Redux | 94.5% | Alibaba, Qwen3.5-397B-A17B model cardchecked 2026-10-15 | - |
| C-Eval | 94% | Alibaba, Qwen3.5-397B-A17B model cardchecked 2026-10-15 | - |
| SuperGPQA | 69.2% | Alibaba, Qwen3.5-397B-A17B model cardchecked 2026-10-15 | - |
| SimpleQA Verified | 36.9% | Thinking Machines, Thinking Machines Inkling model cardchecked 2026-10-15 | - |
| AA-Omniscience Index | -8 | Thinking Machines, Thinking Machines Inkling model cardchecked 2026-10-15 | - |
Code Generation
| Test | Value | Printed by | Setup as printed |
|---|---|---|---|
| LiveCodeBench v6 | 85% | Moonshot AI, Kimi K2.5 model cardchecked 2026-10-15; printed on 2 more pages | - |
| OJBench (C++) | 57.4% | Moonshot AI, Kimi K2.5 model cardchecked 2026-10-15 | - |
| OJBench (Python) | 54.7% | Moonshot AI, Kimi K2.6 model cardchecked 2026-10-15 | - |
Tool Use and Workflows
| Test | Value | Printed by | Setup as printed |
|---|---|---|---|
| Toolathlon | 27.8% | Moonshot AI, Kimi K2.6 model cardchecked 2026-10-15; printed on 3 more pages | - |
| τ³-bench Banking | 14.9% | Mistral AI, Mistral Medium 3.5 model cardchecked 2026-10-15; printed on 1 more page | Effort: max; Harness: gpt-5.2 user simulator (low), 4 trials; terminal- or embedding-based agentic search retrieval, highest score |
| MCP Atlas | 64% | Thinking Machines, Thinking Machines Inkling model cardchecked 2026-10-15 | - |
| MCP Atlas (public set) | 63.8% | Z.ai, GLM-5 model cardchecked 2026-10-15; printed on 1 more page | - |
| MCPMark | 29.5% | Moonshot AI, Kimi K2.6 model cardchecked 2026-10-15; printed on 1 more page | - |
| τ²-bench (domain not stated) | 77% | Alibaba, Qwen3.5-397B-A17B model cardchecked 2026-10-15; printed on 1 more page | - |
| Claw-Eval pass^3 | 52.3% | Moonshot AI, Kimi K2.6 model cardchecked 2026-10-15 | - |
| DeepPlanning | 14.5% | Alibaba, Qwen3.5-397B-A17B model cardchecked 2026-10-15 | - |
| VitaBench | 41.9% | Alibaba, Qwen3.5-397B-A17B model cardchecked 2026-10-15 | - |
| τ³-bench (domain not stated) | 66% | Z.ai, GLM-5.1 model cardchecked 2026-10-15 | - |
| τ³-bench Airline | 76.5% | Mistral AI, Mistral Medium 3.5 model cardchecked 2026-10-15 | Effort: max; Harness: gpt-5.2 user simulator (low), 4 trials; Sierra-reported for Claude Sonnet 4.5 and Qwen3.5 |
| τ³-bench Retail | 72.8% | Mistral AI, Mistral Medium 3.5 model cardchecked 2026-10-15 | Effort: max; Harness: gpt-5.2 user simulator (low), 4 trials; Sierra-reported for Claude Sonnet 4.5 and Qwen3.5 |
| τ³-bench Telecom | 86.8% | Mistral AI, Mistral Medium 3.5 model cardchecked 2026-10-15 | Effort: max; Harness: gpt-5.2 user simulator (low), 4 trials; Sierra-reported for Claude Sonnet 4.5 and Qwen3.5 |
| BFCL v4 | 68.3% | Alibaba, Qwen3.5-397B-A17B model cardchecked 2026-10-15 | - |
| Claw-Eval pass@3 | 75.4% | Moonshot AI, Kimi K2.6 model cardchecked 2026-10-15 | - |
Web Search and Research
| Test | Value | Printed by | Setup as printed |
|---|---|---|---|
| BrowseComp | 74.9% | Moonshot AI, Kimi K2.5 model cardchecked 2026-10-15; printed on 6 more pages | Tools: browsing with context management |
| 60.6% | Moonshot AI, Kimi K2.5 model cardchecked 2026-10-15 | Tools: browsing | |
| 78.4% | Moonshot AI, Kimi K2.5 model cardchecked 2026-10-15 | Tools: browsing, agent swarm | |
| BrowseComp-ZH | 62.3% | Z.ai, GLM-5 model cardchecked 2026-10-15 | Tools: browsing |
| WideSearch (metric not stated) | 72.7 | Alibaba, Qwen3.5-397B-A17B model cardchecked 2026-10-15 | Tools: enabled; search; Context: 256000 |
| Seal-0 | 57.4% | Moonshot AI, Kimi K2.5 model cardchecked 2026-10-15; printed on 1 more page | Tools: search, code interpreter, browsing |
| DeepSearchQA (accuracy) | 77.1% | Moonshot AI, Kimi K2.5 model cardchecked 2026-10-15; printed on 1 more page | Tools: search, code interpreter, browsing |
| DeepSearchQA (F1) | 89% | Moonshot AI, Kimi K2.6 model cardchecked 2026-10-15 | Tools: search, code interpreter, browsing |
| WideSearch (item-F1) | 79% | Moonshot AI, Kimi K2.5 model cardchecked 2026-10-15; printed on 1 more page | Tools: search, code interpreter, browsing, agent swarm |
| 72.7% | Moonshot AI, Kimi K2.5 model cardchecked 2026-10-15 | Tools: search, code interpreter, browsing | |
| FinSearchComp T2 and T3 | 67.8% | Moonshot AI, Kimi K2.5 model cardchecked 2026-10-15 | Tools: search, code interpreter, browsing |
Instruction Following
| Test | Value | Printed by | Setup as printed |
|---|---|---|---|
| IFBench | 70.2 | Alibaba, Qwen3.5-397B-A17B model cardchecked 2026-10-15; printed on 2 more pages | - |
| MultiChallenge | 62.7% | Alibaba, Qwen3.5-397B-A17B model cardchecked 2026-10-15 | - |
| IFEval | 93.9% | Alibaba, Qwen3.5-397B-A17B model cardchecked 2026-10-15 | - |
| COLLIE | 87.8% | Mistral AI, Mistral Medium 3.5 model cardchecked 2026-10-15 | Effort: max |
Multilingual
| Test | Value | Printed by | Setup as printed |
|---|---|---|---|
| MMMLU | 86% | Alibaba, Qwen3.5-397B-A17B model cardchecked 2026-10-15 | - |
| MMLU-ProX | 82.3% | Alibaba, Qwen3.5-397B-A17B model cardchecked 2026-10-15 | - |
| WMT24++ (XCOMET-XXL) | 77.6 | Alibaba, Qwen3.5-397B-A17B model cardchecked 2026-10-15 | - |
| Global PIQA | 89.3% | Alibaba, Qwen3.5-397B-A17B model cardchecked 2026-10-15 | - |
| INCLUDE | 83.3% | Alibaba, Qwen3.5-397B-A17B model cardchecked 2026-10-15 | - |
| MaXIFE | 72.8% | Alibaba, Qwen3.5-397B-A17B model cardchecked 2026-10-15 | - |
| NOVA-63 | 56% | Alibaba, Qwen3.5-397B-A17B model cardchecked 2026-10-15 | - |
| Global-MMLU-Lite | 84% | Thinking Machines, Thinking Machines Inkling model cardchecked 2026-10-15 | - |
Healthcare
| Test | Value | Printed by | Setup as printed |
|---|---|---|---|
| MedXpertQA MM | 65.3% | Alibaba, Qwen3.5-397B-A17B model cardchecked 2026-10-15 | - |
| PMC-VQA | 63.3% | Alibaba, Qwen3.5-397B-A17B model cardchecked 2026-10-15 | - |
| SLAKE | 81.6% | Alibaba, Qwen3.5-397B-A17B model cardchecked 2026-10-15 | - |
Safety and Alignment
| Test | Value | Printed by | Setup as printed |
|---|---|---|---|
| FORTRESS (Adversarial) | 54.1% | Thinking Machines, Thinking Machines Inkling model cardchecked 2026-10-15 | - |
| FORTRESS (Benign) | 98.3% | Thinking Machines, Thinking Machines Inkling model cardchecked 2026-10-15 | - |
| StrongREJECT | 99.5% | Thinking Machines, Thinking Machines Inkling model cardchecked 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.