Codex raises WebSocket request limit 8x to 128MB, making large turns 5x faster
charliermarsh · x · 2026-11-21
Charlie Marsh (creator of ruff) shares that Codex uses a WebSocket for model requests to send data incrementally. The team raised the WebSocket size limit from 16MB to 128MB (8x) so large turns avoid falling back to HTTP, and requests that do exceed the limit are now handled 5x faster. The server also immediately signals oversized requests, skipping pointless upload-and-retry cycles, with similar fast-fail rules applied to other known-hopeless HTTP status codes. Net effect: Codex is significantly faster and more efficient on large turns.
Related event: Codex raises WebSocket limit to 128MB, cutting HTTP fallback ~20%(5 posts)→
More from coding & agent
- CapCut ships ChatGPT plugin for prompt-to-video with editable timeline via MCP — HeyZoyaKhan · 2026-11-21
- ENZO open-source self-hosted AI workspace unifies free-tier APIs and local models — anonymous20256 · 2026-11-21
- Codex raises WebSocket limit to 1 GB, big speed gains on large turns — charliermarsh · 2026-11-21
- Neon boosts Antigravity cache hit rate ~4%, raising quota effectiveness by at least 20% — cgarciae88 · 2026-11-21
- ArcFront launches with 4 of 5 frontier labs as customers, targeting agent evals — eptwts · 2026-11-21
- PostHog engineer on moving into AI product engineering: learning evals and agent interfaces — rseroter · 2026-11-21