KwaiKAT Team Releases KAT-Coder-V2.5: An Agentic Coding Model Trained on 100,000+ Verifiable Repository Environments⭐8

KwaiKAT Team launches KAT-Coder-V2.5, an agentic AI trained on 100K+ verifiable repositories, achieving state-of-the-art code generation and debugging with 72% ...

DeepSeek Releases DSpark: A Speculative Decoding Framework That Accelerates DeepSeek-V4 Per-User Generation by 60–85% Over MTP-1⭐7

DeepSeek releases DSpark, a speculative decoding framework boosting DeepSeek-V4 per-user generation speed by 60–85% over MTP-1, with open-source availability.

Z.ai Launches GLM-5.2: 1M-Token Context, Two Thinking-Effort Levels, and No Launch Benchmarks⭐8

Z.ai launches GLM-5.2 with a usable 1M-token context and two thinking-effort levels, but no benchmarks disclosed at launch, sparking industry curiosity and skep...

Moonshot AI Unveils Kimi K2.7-Code: A Coding Model with +21.8% Improvement on Kimi Code Bench v2 Over K2.6⭐7

Moonshot AI launches Kimi K2.7-Code, a coding LLM with a 21.8% improvement over K2.6 on Kimi Code Bench v2, boosting code generation and debugging.