LLM+ Advances: Million-Token Contexts, Cheaper Models
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- OpenAI expands its enterprise coding suite, rolling out Codex updates that target Fortune 500 developers and promise tighter integration with internal tools.
- DeepSeek unveiled a text‑in‑image encoding method that cuts LLM compute cost by encoding language into visual tokens, a novel efficiency hack.
- MIT CSAIL released a recursive LLM architecture that splits inputs into chunks processed by multiple model copies, enabling reliable handling of up to a million tokens.
- Codex gains multi‑file codebase support, allowing developers to edit and test large projects in a single session.
Why it matters: Not enough in the article to say why this matters.
