OpenAI explains why Codex defaults to 300k context instead of 1M
prd_008 · x · 2026-10-10
Thibault Sottiaux, OpenAI's Codex engineering lead, explained why Codex uses a 300k context window: after running the numbers, they found it performs better. The model can technically run at 1M context, but that would consume more of users' usage allowance — "good defaults are important," he noted, revealing a deliberate trade-off between context length and usage cost.
More from coding & agent
- Andrew Ng releases 1-hour course on building agentic knowledge graphs from scratch — nikola_mr64990 · 2026-10-10
- Mnemos: an engram-inspired memory architecture doubles short-term recall vs Claude's native memory — RileyRalmuto · 2026-10-10
- LLM reliability #6-7: guardrails, fallbacks, and production monitoring — goyalshaliniuk · 2026-10-10
- LLM reliability #4-5: abstaining gracefully and controlling context quality — goyalshaliniuk · 2026-10-10
- LLM reliability #3-4: strong eval suites and graceful abstention — goyalshaliniuk · 2026-10-10
- 7 ways to improve LLM reliability, from RAG grounding to production monitoring — goyalshaliniuk · 2026-10-10