BenchMark'd
No ratings yet
Be the first to review this model
The first model to integrate thinking directly into tool use, featuring DeepSeek Sparse Attention for efficient long-context processing. Achieves gold-medal performance in the 2025 IMO and IOI, performing comparably to GPT-5. A frontier-class open model pushing the boundaries of what open-weight AI can achieve.
Released
December 1, 2025
Parameters
685B (MoE, 37B active)
Context
164K
Pricing
Open Source
Last updated: March 15, 2026
Benchmark scores may vary based on evaluation methodology and conditions.