Beyond rate limits: scaling access to Codex and Sora

Source: OpenAI Date: 2026-02-13 URL: https://openai.com/index/beyond-rate-limits

Summary

OpenAI post from February 2026 announcing expanded access to Codex (the coding agent, not the earlier code-completion model) and Sora — moving beyond the waitlist and rate-limited research preview phases that both products had been in since their respective launches. The expansion likely included higher usage limits for API customers, broader geographic availability, and potentially new pricing tiers. Both products had faced significant demand pressure with constrained access through late 2025.

Implications

Infrastructure readiness as a competitive signal. Sora’s January 2024 demo was one of the most viral AI moments of that year; that it took until early 2026 to fully scale access indicates the inference costs and safety review processes were genuinely constraining, not marketing scarcity. Codex’s constrained access similarly reflected the agent infrastructure immaturity more than demand management.

Codex-as-product, not API. The pairing of Codex and Sora in a single scaling announcement is interesting — these are OpenAI’s two “showpiece” products that require the most compute and have the most safety review overhead. Scaling them together suggests infrastructure economies were being realized across both.

Thread: OpenAI product access. Sits alongside the ChatGPT Go launch (January 2026) and the edu-for-countries rollout — OpenAI’s broader Q1 2026 access-expansion push after the infrastructure investment from Stargate started paying off.

Watch: Whether expanded Codex access accelerates adoption in the enterprise coding segment, where Cursor and GitHub Copilot have been faster to serve demand.