Microsoft AI MAI-Thinking-1 69.0% microsoft.ai for MAI-Thinking-1 on IFBench Original (opens in a new tab) Accuracy Reported August 12, 2026 Technical report Microsoft AI Self-reported Microsoft Record for MAI-Thinking-1 on IFBench Original 70003 Meta AI (originally Facebook AI Research) Muse Glimmer 30B (high)77.0% dev.meta.ai for Muse Glimmer 30B (high) on IFBench Original (opens in a new tab) Accuracy Reported August 10, 2026 Model page Meta Self-reported Record for Muse Glimmer 30B on IFBench Original 80004 Alibaba (Tongyi) Qwen3.8-Max 82.8% huggingface.co for Qwen3.8-Max on IFBench Original (opens in a new tab) Accuracy Reported August 2, 2026 Model card Qwen Self-reported Qwen Team Record for Qwen3.8-Max on IFBench Original 130008 Thinking Machines Lab Inkling-Small (effort=0.99)82.2% thinkingmachines.ai for Inkling-Small (effort=0.99) on IFBench Original (opens in a new tab) Accuracy Reported July 30, 2026 Launch post Thinking Machines Lab Self-reported Record for Inkling-Small on IFBench Original 160002 Google (DeepMind) Gemma 4 26B A4B Instruct (thinking mode)72.0% arxiv.org for Gemma 4 26B A4B Instruct (thinking mode) on IFBench Original (opens in a new tab) Accuracy Reported July 24, 2026 Technical report Google DeepMind Self-reported Record for Gemma 4 26B A4B Instruct on IFBench Original 35001 Google (DeepMind) Gemma 4 31B Instruct (thinking mode)76.0% arxiv.org for Gemma 4 31B Instruct (thinking mode) on IFBench Original (opens in a new tab) Accuracy Reported July 24, 2026 Technical report Google DeepMind Self-reported Record for Gemma 4 31B Instruct on IFBench Original 35002 Google (DeepMind) Gemma 4 E2B Instruct (thinking mode)38.0% arxiv.org for Gemma 4 E2B Instruct (thinking mode) on IFBench Original (opens in a new tab) Accuracy Reported July 24, 2026 Technical report Google DeepMind Self-reported Record for Gemma 4 E2B Instruct on IFBench Original 35003 Google (DeepMind) Gemma 4 E4B Instruct (thinking mode)44.0% arxiv.org for Gemma 4 E4B Instruct (thinking mode) on IFBench Original (opens in a new tab) Accuracy Reported July 24, 2026 Technical report Google DeepMind Self-reported Record for Gemma 4 E4B Instruct on IFBench Original 35004 Google (DeepMind) Gemma 4 12B Instruct (thinking mode)74.0% arxiv.org for Gemma 4 12B Instruct (thinking mode) on IFBench Original (opens in a new tab) Accuracy Reported July 24, 2026 Technical report Google DeepMind Self-reported Record for Gemma 4 12B Instruct on IFBench Original 35005 Thinking Machines Lab Inkling (effort=0.99)79.8% thinkingmachines.ai for Inkling (effort=0.99) on IFBench Original (opens in a new tab) Accuracy Reported July 15, 2026 Launch post Thinking Machines Lab Self-reported Record for Inkling on IFBench Original 160001 Alibaba (Tongyi) Qwen3.7-Plus 79.1% alibabacloud.com for Qwen3.7-Plus on IFBench Original (opens in a new tab) Accuracy Reported June 3, 2026 Launch post Alibaba Cloud Community Self-reported Qwen Team Record for Qwen3.7-Plus on IFBench Original 130007 Mistral AI Mistral Medium 3.5 (high)68.0% huggingface.co for Mistral Medium 3.5 (high) on IFBench Original (opens in a new tab) Accuracy Reported May 22, 2026 Model card Mistral AI Self-reported Record for Mistral Medium 3.5 on IFBench Original 90005 Alibaba (Tongyi) Qwen3.7-Max 79.1% alibabacloud.com for Qwen3.7-Max on IFBench Original (opens in a new tab) Accuracy Reported May 21, 2026 Launch post Alibaba Cloud Community Self-reported Qwen Team Record for Qwen3.7-Max on IFBench Original 130006 Alibaba (Tongyi) Qwen3.6-Plus 74.2% alibabacloud.com for Qwen3.6-Plus on IFBench Original (opens in a new tab) Accuracy Reported April 2, 2026 Launch post Alibaba Cloud Community Self-reported Qwen Team Record for Qwen3.6-Plus on IFBench Original 130004 Mistral AI Mistral Small 4 (high)48.0% huggingface.co for Mistral Small 4 (high) on IFBench Original (opens in a new tab) Accuracy Reported March 16, 2026 Model card Mistral AI Self-reported Record for Mistral Small 4 on IFBench Original 90004 Mistral AI Mistral Small 4 (none)35.7% huggingface.co for Mistral Small 4 (none) on IFBench Original (opens in a new tab) Accuracy Reported March 16, 2026 Model card Mistral AI Self-reported Record for Mistral Small 4 on IFBench Original 90004 Alibaba (Tongyi) Qwen3.5-397B-A17B 76.5% alibabacloud.com for Qwen3.5-397B-A17B on IFBench Original (opens in a new tab) Accuracy Reported February 17, 2026 Launch post Alibaba Cloud Community Self-reported Qwen Team Record for Qwen3.5-397B-A17B on IFBench Original 130003 Alibaba (Tongyi) Qwen3-Max-Thinking 70.9% alibabacloud.com for Qwen3-Max-Thinking on IFBench Original (opens in a new tab) Accuracy Reported January 25, 2026 Launch post Alibaba Cloud Community Self-reported Qwen Team Record for Qwen3-Max-Thinking on IFBench Original 130002