The Brief

New briefs daily around 7 AM Eastern

OpenAI's chief scientist says reasoning models can now self-improve

OpenAI's chief scientist argues reasoning models have crossed into capability jumps that could sustain recursive self-improvement, and says monitoring and coordination must strengthen before the next jump lands.

• 01Read

  OpenAI's per-researcher AI spend jumped before GPT-6 Astra

  A look at how OpenAI's research team leans on coding agents internally, anchored by a chart showing AI spend per researcher jumping sharply in late July, around when GPT-6 Astra reportedly reached internal users first.

  simonwillison.net→
• 02Read

  Tencent releases EVIE-8B, tops the ViDoRe V3 benchmark

  Tencent released EVIE-8B, a 4096-dimensional late-interaction visual document retrieval model that tops the ViDoRe V3 leaderboard and serves as the distillation teacher for the smaller EVIE-4.5B.

  Tencent model releases→
• 03Read

  Tencent's EVIE-4.5B shrinks retrieval indexes 20x

  Tencent open-sourced EVIE-4.5B, a late-interaction visual document retrieval model with elastic embeddings from 64 to 2048 dimensions and a training-free compression scheme that cuts index size roughly 20x while still leading ViDoRe V3.

  Tencent model releases→
• 04Read

  Figma's CTO says AI commoditizes code, not judgment

  Figma's CTO argues AI is commoditizing code translation, from design to code and between languages, but not the harder work of choosing problems, building abstractions, and reasoning about systems from first principles.

  Figma Blog→