--- title: "MAI-Thinking-1, MAI-Code-1-Flash" date: 2026-06-03 url: https://blog.duyet.net/note/mai --- | **Benchmark** | **MAI-Thinking-1** | **MAI-Code-1-Flash** | **Claude Haiku 4.5** | **Sonnet 4.6** | **Opus 4.6** | **Kimi K2.6** | **GLM-5.1** | | ---------------------- | -----------------: | -------------------: | --------------------------: | -------------------------: | -------------------------: | -----------------------------: | ---------------------------------------: | | SWE-Bench Pro | 52.8 | 51.2 | 35.2 | - | 53.4 | **58.6** | 58.4 | | SWE-Bench Verified | 73.5 | 71.6 | 66.6 | 79.6 | **80.8** | 80.2 | - | | SWE-Bench Multilingual | - | 65.5 | 62.7 | 75.9 | **77.8** | 76.7 | - | | AIME 2025 | 97.0 | - | 80.7 no tools / 96.3 Python | 95.6 | **99.8** | - | - | | AIME 2026 | 94.5 | 92.5 | 83.3 | - | - | **96.4** | 95.3 | | HMMT Feb 2026 | 84.9 | - | - | - | - | **92.7** | 82.6 | | GPQA Diamond | 84.2 | 84.6 | 73.2 | 89.9 | **91.3** | 90.5 | 86.2 | | LCB v6 | 87.7 | - | - | - | - | **89.6** | - | | AMO Bench | - | **40.0** | 16.0 | - | - | - | - | | Frontier Math | - | **6.3** | 2.8 | - | - | - | - | | HLE | - | 18.0 | 9.5 | 33.2 no tools / 49.0 tools | 40.0 no tools / 53.0 tools | 34.7 no tools / **54.0 tools** | 31.0 no tools / 52.3 tools | | Frontier Science | - | **58.2** | 42.3 | - | - | - | - | | Artifacts Bench | - | 36.4 | **36.6** | - | - | - | - | | Terminal-Bench 2.0 | 46.0 | 54.8 | 41.6 | 59.1 | 65.4 | 66.7 | 63.5 Terminus-2 / **69.0 self-reported** | | IF Bench | - | **75.0** | 46.1 | - | - | - | - | | Advanced IF | - | **71.4** | 56.9 | - | - | - | - | | Robust IF Bench | - | **61.2** | 45.0 | - | - | - | - | | τ²-Bench | - | **71.7** | 54.7 | - | - | - | - | ## Sources - Microsoft AI, **MAI-Thinking-1 technical report**, for MAI-Thinking-1 scores and several comparison values. ([Microsoft AI](https://microsoft.ai/wp-content/uploads/2026/06/main_20260602_2.pdf?utm_source=chatgpt.com "MAI-Thinking-1: Building a Hill-Climbing Machine - Microsoft AI")) - Microsoft AI, **Introducing MAI-Thinking-1**, for headline AIME 2025 and AIME 2026 values. ([Microsoft AI](https://microsoft.ai/news/introducing-mai-thinking-1/?utm_source=chatgpt.com "Introducing MAI-Thinking-1 | Microsoft AI") - Microsoft AI, **Introducing MAI-Code-1-Flash**, for MAI-Code-1-Flash vs Claude Haiku 4.5 benchmark values. ([Microsoft AI](https://microsoft.ai/news/introducingmai-code-1-flash/?utm_source=chatgpt.com "Introducing MAI-Code-1-Flash - Microsoft AI")) - Anthropic, **Claude Haiku 4.5 announcement**, for official Haiku context and Anthropic-reported SWE-Bench Verified score caveat. ([Anthropic](https://www.anthropic.com/news/claude-haiku-4-5?utm_source=chatgpt.com "Introducing Claude Haiku 4.5")) - Anthropic, **Claude Sonnet 4.6 System Card**, for Sonnet 4.6 SWE-Bench Verified and SWE-Bench Multilingual values. ([Anthropic](https://anthropic.com/claude-sonnet-4-6-system-card?utm_source=chatgpt.com "Claude Sonnet 4.6 System Card")) - Anthropic, **Claude Opus 4.6 System Card**, for Opus 4.6 SWE-Bench Verified and SWE-Bench Multilingual values. ([Anthropic](https://www-cdn.anthropic.com/0dd865075ad3132672ee0ab40b05a53f14cf5288.pdf?utm_source=chatgpt.com "Claude Opus 4.6 System Card")) - Moonshot AI, **Kimi K2.6 Tech Blog / Model Card**, for Kimi K2.6 benchmark values. ([Kimi](https://www.kimi.com/blog/kimi-k2-6?utm_source=chatgpt.com "Kimi K2.6 Tech Blog: Advancing Open-Source Coding")) - Z.ai, **GLM-5.1: Towards Long-Horizon Tasks**, and ModelScope GLM-5.1 card, for GLM-5.1 values. ([Z.ai](https://z.ai/blog/glm-5.1?utm_source=chatgpt.com "GLM-5.1: Towards Long-Horizon Tasks")) - Terminal-Bench official leaderboard, for Terminal-Bench 2.0 reference leaderboard context. ([tbench.ai](https://www.tbench.ai/leaderboard/terminal-bench/2.0?utm_source=chatgpt.com "terminal-bench@2.0 Leaderboard"))