OpenAI launches $100 Pro tier to plug Codex gap as Anthropic gains ground

Product change: new $100 Pro tier and pay‑as‑you‑go seats

OpenAI said it is adding a $100-per-month Pro subscription tier to better serve heavy users of Codex, its AI coding agent. The new tier gives five times the Codex usage allowance of the existing Plus plan and preserves Pro‑level features such as exclusive Pro models and unlimited Instant and Thinking model calls. OpenAI also extended a promotion that doubles Codex allowances for existing $200 Pro subscribers through May 31 and has raised Codex rate limits; Business and Enterprise workspaces were already offered dedicated Codex seats on a pay‑as‑you‑go basis to make pilots and scaling cheaper for teams.

Market context: developer flows and fierce commercial rivalry

The move addresses a commercial pain point: many developers found Plus insufficient and the $200 Pro too expensive, and reportedly migrated to Anthropic’s Claude Code. OpenAI’s Codex engineering lead Thibault Sottiaux said Codex weekly users have reached about 3 million, up from 2 million a month earlier. It has been reported that Codex’s annualized revenue surpassed $1 billion in January 2026. Meanwhile, it has been reported that Anthropic announced annualized recurring revenue (ARR) above $30 billion, versus around $25 billion for OpenAI by some industry counts — a gap that has intensified pricing and product adjustments.

Why it matters: pricing, tokens and the battle for developers

This is about economics as much as models. Token‑hungry programming workflows have driven fresh monetization models across the industry: layered subscriptions, pay‑per‑use, and outcome‑based pricing. Investors and operators expect top models to command higher token prices while simpler workloads move toward low‑cost or ad‑supported tiers. Can a new price point stop developer attrition when model performance — not just cost — is the decisive factor? For OpenAI, the $100 tier is a defensive, pragmatic fix. But it has been reported that Anthropic has already built substantial B‑to‑B momentum in coding, and price tweaks alone may not be enough to change developer preferences amid broader regulatory and geopolitical scrutiny of AI supply chains.