State of AI Pricing: October 2026

The month so far

October's big price move came at the bottom of the market. On 7 October Anthropic released Claude Haiku 5.5 at $0.10 / $0.50 per million tokens, 90% below Haiku 4.5's $1 / $5 and the same per-token price as OpenAI's GPT-6 Luna. There is a catch: a request whose prompt is over 100,000 tokens is billed at five times the rate. Full Haiku 5.5 pricing.

The same day, Anthropic halved cached input on Claude Sonnet 5.5 from $0.20 to $0.10 per million tokens. With GPT-6.1 Sol's cached input cut to $0.10 at its launch on 29 September, OpenAI’s and Anthropic’s main mid-priced models now cost the same: $2 / $10, with cached input at $0.10. For agents, which resend the same context on every step, the cache price matters more than the headline rate: for a coding agent using 750M input tokens a month, 85% of it cached, Sonnet 5.5 now costs about $369 a month; at the old cache price it would have been about 17% more. Anthropic puts the saving on most agent workloads at about 20%.

Hosted open-weight models got cheaper too. By 13 October, when we last checked, the cheapest hosts listed on OpenRouter had repriced six of the open models we track, mostly downwards, led by gpt-oss-120b at $0.030 / $0.15, down from $0.15 / $0.60 on 27 September.

Frontier prices didn't move. No frontier model launched or changed price in the first two weeks of October, after a September in which Claude Opus 5.5, GPT-6 Astra, Grok 4.7 and Muse Spark 1.3 all arrived.

Subscriptions and coding plans

ChatGPT Pro is now sold as three plans, at $100, $200 and $500 a month; the $500 plan adds an Ultrafast mode in Codex. Claude Max subscribers get monthly credits for the Claude API from October, $100 on Max 5x and $200 on Max 20x, though Anthropic says the credits can't be spent in Claude Code. Our new coding plans comparison has every tier of Claude Code, Codex, Cursor and GitHub Copilot, with the API cost of the same work.

What to watch for the rest of October

• 14 October: GPT-5.5 leaves ChatGPT and Codex on every plan. The API version isn't affected.
• 23 October: OpenAI shuts down o1 and the original gpt-4o snapshot in the API.
• Later this year: GPT-5.6 Sol's promotional price is guaranteed until at least 21 November, and Gemini 3.8 Flash's promotional price ends on 31 December, when it rises to $1.50 / $7.50.