BenchMark'd
No ratings yet
Be the first to review this model
Alibaba's flagship hybrid reasoning model that dynamically switches between thinking and non-thinking modes. Trained on 36 trillion tokens, double its predecessor, with both dense and sparse MoE variants available. Sets new benchmarks for open-source AI in coding, math, and multilingual reasoning.
Released
April 28, 2025
Parameters
235B (MoE, 22B active)
Context
128K
Pricing
Open Source
Last updated: March 15, 2026
Benchmark scores may vary based on evaluation methodology and conditions.