| MAI-Thinking-1 | Riemann-bench October 2026 leaderboard | 4.8% | surgehq.ai for MAI-Thinking-1 on Riemann-bench October 2026 leaderboard (opens in a new tab)Archived copy (opens in a new tab)Riemann-bench Pass@1Reported Benchmark leaderboardSurge AIIndependentRecord for MAI-Thinking-1 on Riemann-bench October 2026 leaderboard | 70003 |
| MAI-Thinking-1 | SWE-bench Verified | 73.5% | microsoft.ai for MAI-Thinking-1 on SWE-bench Verified (opens in a new tab)Resolved TasksReported Technical reportMicrosoft AISelf-reportedMicrosoftRecord for MAI-Thinking-1 on SWE-bench Verified | 70003 |
| MAI-Thinking-1 | SWE-bench Pro Public | 52.8% | microsoft.ai for MAI-Thinking-1 on SWE-bench Pro Public (opens in a new tab)Public tasks resolvedReported Technical reportMicrosoft AISelf-reportedMicrosoftRecord for MAI-Thinking-1 on SWE-bench Pro Public | 70003 |
| MAI-Thinking-1 | SimpleQA Verified Original | 31.0% | microsoft.ai for MAI-Thinking-1 on SimpleQA Verified Original (opens in a new tab)AccuracyReported Technical reportMicrosoft AISelf-reportedMicrosoftRecord for MAI-Thinking-1 on SimpleQA Verified Original | 70003 |
| MAI-Thinking-1 | GPQA Diamond | 84.2% | microsoft.ai for MAI-Thinking-1 on Graduate-Level Google-Proof Q&A Diamond (opens in a new tab)GPQA Diamond AccuracyReported Technical reportMicrosoft AISelf-reportedMicrosoftRecord for MAI-Thinking-1 on Graduate-Level Google-Proof Q&A Diamond | 70003 |
| MAI-Thinking-1 | IFBench Original | 69.0% | microsoft.ai for MAI-Thinking-1 on IFBench Original (opens in a new tab)AccuracyReported Technical reportMicrosoft AISelf-reportedMicrosoftRecord for MAI-Thinking-1 on IFBench Original | 70003 |
| MAI-Thinking-1 | Terminal-Bench 2.0 | 46.0% | microsoft.ai for MAI-Thinking-1 on Terminal-Bench 2.0 (opens in a new tab)Terminal-Bench AccuracyReported Technical reportMicrosoft AISelf-reportedMicrosoftRecord for MAI-Thinking-1 on Terminal-Bench 2.0 | 70003 |
| MAI-Thinking-1 | MMLU-Pro Original | 85.0% | microsoft.ai for MAI-Thinking-1 on MMLU-Pro Original (opens in a new tab)AccuracyReported Technical reportMicrosoft AISelf-reportedMicrosoftRecord for MAI-Thinking-1 on MMLU-Pro Original | 70003 |
| MAI-Code-1-Flash (adaptive thinking) | SWE-bench Multilingual — GitHub Copilot production harness | 65.5% | microsoft.ai for MAI-Code-1-Flash (adaptive thinking) on SWE-bench Multilingual — GitHub Copilot production harness (opens in a new tab)Resolved TasksReported Launch postMicrosoft AISelf-reportedMicrosoftRecord for MAI-Code-1-Flash on SWE-bench Multilingual — GitHub Copilot production harness | 70002 |
| MAI-Code-1-Flash (adaptive thinking) | SWE-bench Verified — GitHub Copilot production harness | 71.6% | microsoft.ai for MAI-Code-1-Flash (adaptive thinking) on SWE-bench Verified — GitHub Copilot production harness (opens in a new tab)Resolved TasksReported Launch postMicrosoft AISelf-reportedMicrosoftRecord for MAI-Code-1-Flash on SWE-bench Verified — GitHub Copilot production harness | 70002 |
| MAI-Code-1-Flash (adaptive thinking) | SWE-bench Pro Public — GitHub Copilot production harness | 51.2% | microsoft.ai for MAI-Code-1-Flash (adaptive thinking) on SWE-bench Pro Public — GitHub Copilot production harness (opens in a new tab)Public tasks resolvedReported Launch postMicrosoft AISelf-reportedMicrosoftRecord for MAI-Code-1-Flash on SWE-bench Pro Public — GitHub Copilot production harness | 70002 |
| MAI-Code-1-Flash | GPQA Diamond | 84.6% | microsoft.ai for MAI-Code-1-Flash on Graduate-Level Google-Proof Q&A Diamond (opens in a new tab)GPQA Diamond AccuracyReported Launch postMicrosoft AISelf-reportedMicrosoftRecord for MAI-Code-1-Flash on Graduate-Level Google-Proof Q&A Diamond | 70002 |
| MAI-Code-1-Flash (adaptive thinking) | Terminal-Bench 2.0 — GitHub Copilot production harness | 54.8% | microsoft.ai for MAI-Code-1-Flash (adaptive thinking) on Terminal-Bench 2.0 — GitHub Copilot production harness (opens in a new tab)Terminal-Bench AccuracyReported Launch postMicrosoft AISelf-reportedMicrosoftRecord for MAI-Code-1-Flash on Terminal-Bench 2.0 — GitHub Copilot production harness | 70002 |
| Phi-4 Reasoning Vision 15B (automatic think/nothink) | MathVerse MINI | 44.9% | huggingface.co for Phi-4 Reasoning Vision 15B (automatic think/nothink) on MathVerse MINI (opens in a new tab)AccuracyReported Model cardMicrosoftSelf-reportedRecord for Phi-4 Reasoning Vision 15B on MathVerse MINI | 70001 |
| Phi-4 Reasoning Vision 15B (automatic think/nothink) | MMStar Original | 64.5% | huggingface.co for Phi-4 Reasoning Vision 15B (automatic think/nothink) on MMStar Original (opens in a new tab)AccuracyReported Model cardMicrosoftSelf-reportedRecord for Phi-4 Reasoning Vision 15B on MMStar Original | 70001 |
| Phi-4 Reasoning Vision 15B (automatic think/nothink) | AI2D TEST | 84.8% | huggingface.co for Phi-4 Reasoning Vision 15B (automatic think/nothink) on AI2D TEST (opens in a new tab)AccuracyReported Model cardMicrosoftSelf-reportedRecord for Phi-4 Reasoning Vision 15B on AI2D TEST | 70001 |
| Phi-4 Reasoning Vision 15B (automatic think/nothink) | HallusionBench Original | 64.4% | huggingface.co for Phi-4 Reasoning Vision 15B (automatic think/nothink) on HallusionBench Original (opens in a new tab)AccuracyReported Model cardMicrosoftSelf-reportedRecord for Phi-4 Reasoning Vision 15B on HallusionBench Original | 70001 |
| Phi-4 Reasoning Vision 15B (automatic think/nothink) | ChartQA TEST | 83.3% | huggingface.co for Phi-4 Reasoning Vision 15B (automatic think/nothink) on ChartQA TEST (opens in a new tab)AccuracyReported Model cardMicrosoftSelf-reportedRecord for Phi-4 Reasoning Vision 15B on ChartQA TEST | 70001 |
| Phi-4 Reasoning Vision 15B (automatic think/nothink) | MathVista MINI | 75.2% | huggingface.co for Phi-4 Reasoning Vision 15B (automatic think/nothink) on MathVista MINI (opens in a new tab)AccuracyReported Model cardMicrosoftSelf-reportedRecord for Phi-4 Reasoning Vision 15B on MathVista MINI | 70001 |
| Phi-4 Reasoning Vision 15B (automatic think/nothink) | OCRBench Original | 76.0% | huggingface.co for Phi-4 Reasoning Vision 15B (automatic think/nothink) on OCRBench Original (opens in a new tab)AccuracyReported Model cardMicrosoftSelf-reportedRecord for Phi-4 Reasoning Vision 15B on OCRBench Original | 70001 |