DeepSeek targets 'infinite context' and test-time parametric continual learning
teortaxesTex · x · 2026-09-13
Blogger teortaxesTex reads DeepSeek's roadmap: since V3 the goal has been 'infinite context' — KV per token reduced to 890 bytes, near-linear costs at 1M context, extendable to millions via RLM and harness tricks. He also surfaces a mission statement from DeepSeek's Shengding Hu: 'Not RSI, not harness-level evolving, straight shot to test-time parametric continual learning (similar to SSI, guessed),' with a call for eligible collaborators.
Related event: DeepSeek Researcher Reveals Path to Test-Time Parametric Continual Learning(2 posts)→
More from Models
- DeepSeek V4.1 Flash accused of public benchmark contamination, Kimi K3 possibly too — teortaxesTex · 2026-09-13
- 45% of overnight benchmark rollouts failed mid-turn amid OpenAI capacity issues — dejavucoder · 2026-09-13
- Gemini turns out surprisingly good at translating swarm language to and from English — xeophon · 2026-09-13
- If AGI is here, why are internal models like Fable and Astra still so expensive? — SaW120 · 2026-09-13
- DeepSeek V4.1 costs less to serve but prices 2x higher per output token, margins likely up — teortaxesTex · 2026-09-13
- Migrating from Claude Code to Codex: history, plugins and skills don't carry over — Inevitable2727 · 2026-09-13