OpenAI and Anthropic cut LLM prices by up to 80% since June
Original title: OpenAI and Anthropic cut their LLM prices by up to 80% within a few months. Why?
Cache-read cuts matter most for coding agents, where cache reads are 96–99% of token spend, the author says.
A blog timeline tracks OpenAI and Anthropic price cuts from June through late September 2026 across GPT-5.6, GPT-6, Claude Sonnet, Opus, and Fable lines. OpenAI cut GPT-5.6 Luna by 80% on July 30; later GPT-6.1 Sol matched a 5%-of-input cache-read rate. Anthropic made Sonnet 5’s launch discount permanent and priced Opus 5.5 lower because it needs less compute to serve. The author treats the moves as tit-for-tat commodity pricing, not quality cuts, and cites Vercel gateway data that cost per token fell while aggregate spend rose.
Unconfirmed: The tit-for-tat motive and some quality comparisons are the author’s inference, including personal use of GPT-6.1 Sol.
- GPT-5.6 Luna fell to $0.20 / $1.20 / $0.02 on July 30.
- Fable 5.1 kept $10 / $50 input and output but cut cache reads 75% to $0.25.
- GPT-6 Sol launched at $2 / $10 / $0.20; GPT-6.1 Sol then halved cache reads to $0.10.
- Opus 5.5 is $4 / $20 / $0.20, about 40–60% cheaper overall than Opus 5.
- Vercel’s gateway shows cost per token at 43.3% of the April 2026 level.
- Products
- CodexClaude Code
- Organizations
- Vercel
From the original article
11 min read at minimaxir.com
Let’s look at a quick timeline of the major LLM model announcements from both OpenAI and Anthropic over the past few months. All pricing numbers provided are input MTok / output MTok / cache read MTok . OpenAI announced GPT-5.
By minimaxir · minimaxir.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.