All editions

1. September 11, 2026 · 3 min read

   Edition 012 — DeepSeek shrinks the KV cache, SWE-2 pushes coding's Pareto frontier, and VLMs learn to play robots

   DeepSeek V4.1 Flash compresses the global KV cache to 890 bytes/token, Cognition's SWE-2 pushes the Pareto frontier of agentic coding, and two research papers show VLMs playing robots and multi-agent pipelines optimizing themselves 2.5x faster.

   • robotics
   • DeepSeek
   • V4.1-Flash
   • Cognition
   • SWE-2
   • VLMs
   • multi-agent

2. September 2, 2026 · 3 min read

   Edition 004 — The open stack gets its eyes: DeepSeek vision, Qwen 27B dense, LTX-2.5 sound

   Three releases, one pattern — the open stack is racing to catch the closed frontier on sight, scale, and sound while keeping weights and licenses yours. DeepSeek puts eyes on an MIT agent, Qwen's dense 27B refuses to bloat, and open video finally learned to hear.

   • open-weights
   • DeepSeek
   • multimodal
   • qwen
   • video-generation

Subscribe for each new edition →