1 source 7h ago
OpenAI and Anthropic Cut LLM Prices as Open-Weight Models Take Token Share
OpenAI and Anthropic cut LLM usage fees by up to 80% between July and September 2026.
Reporting from 1 source: GIGAZINE.
Why it matters
Vercel's AI Gateway data showing 48.0% of teams changed their most-used model in September suggests price cuts are a response to an inference market where users can move, not a favor to them.
Key Facts
- OpenAI cut GPT-5.6 Luna prices on July 30, 2026, from 1 dollar to 0.20 dollars input and 6 dollars to 1.20 dollars output per 1 million tokens.
- Opus 5.5's cache read fee is 0.20 dollars per 1 million tokens, a 60% cut from Opus 5.
- At Vercel's AI Gateway, 48.0% of teams changed their most-used model during September, and 28.6% changed the company behind it.
- Data scientist Max Woolf points to low switching costs and open-weight models as reasons for the price cuts.
On July 30, OpenAI reduced GPT-5.6 Luna input from 1 dollar to 0.20 dollars and output from 6 dollars to 1.20 dollars per 1 million tokens. Anthropic's Opus 5.5, announced September 22, costs 4 dollars input and 20 dollars output. Data scientist Max Woolf points to low switching costs and competition from open-weight models.
Anthropic's Opus 5.5 carries a cache read fee of 0.20 dollars per 1 million tokens, down 60% from Opus 5, because Anthropic says cache reads make up most of the cost in agent and coding work. OpenAI's GPT-6.1 Sol, announced September 29, sits at 2 dollars input and 10 dollars output, against 5 and 30 for GPT-5.6 Sol in June.
The migration numbers come from Vercel's AI Gateway, which routes traffic across multiple AI models. It recorded 28.6% of teams changing the company behind their main model over the same month, alongside the 48.0% who changed the model itself.
Written with AI by Yomimono from the 1 cited source below, including Japanese-language reporting where cited. See our editorial policy.
Add Yomimono as a preferred source on Google