Alibaba has officially unveiled Qwen3.8-Max, a 2.4-trillion-parameter Mixture-of-Experts flagship built for long-horizon agent execution and coding, with promises of open weights on the horizon. Meanwhile, mathematical AI reaches a historic benchmark as RedNote AI’s beta model dots-note-3.0 scored a perfect 42/42 on the 2026 International Mathematical Olympiad (IMO), delivering full written proofs with zero human intervention.

We’ll talk about:

• A 2.4T parameter model with a 1M context window claiming performance second only to Claude Fable 5, set to release downloadable open weights.
• dots-note-3.0 solving all six complex Olympiad proof problems using combined natural-language reasoning and Python verification.
• xAI introducing voice consistency, multi-reference image conditioning, native 10 ...