Mistral Mathstral 2.0 sets new open-source SOTA on MATH and GSM8K-Hard
Mathstral 2.0 (32B model specialized for mathematics) achieves 94.1% on MATH benchmark and 98.7% on GSM8K-Hard, surpassing all open models and approaching closed frontier performance through curriculum-based continual pretraining and heavy math token emphasis.
new
Mistral Mathstral Large 2.1 reaches SOTA on advanced symbolic math and proof verification
Anthropic Claude 5.2 Opus executes 28-day autonomous materials science simulation campaign
xAI Grok-6.1 adds native 4D spatiotemporal reasoning for video + sensor fusion
Google DeepMind AlphaZero 4 masters quantum circuit optimization for fault-tolerant computing
OpenAI o4-proto-6 enables 96-hour autonomous ML model fine-tuning and deployment loop
Mistral Pixtral Large 2.5 reaches SOTA open multimodal reasoning on MMMU-Pro and MathVista-Hard
Anthropic Claude 5 Sonnet demonstrates 14-day autonomous scientific literature review + hypothesis generation
xAI Grok-5.1 integrates real-time multi-agent debate for improved truthfulness and calibration
Google DeepMind AlphaFold 4 predicts all known protein complexes with near-experimental accuracy
OpenAI o4-proto-4 achieves 82% autonomous resolution on full SWE-Bench Verified + Multi-File