Analysis: OpenAI and Anthropic cut API prices by up to 80% between July and September 2026, driven by low switching costs and open-weight rivals
Confirmed
Status
- Claim
Confirmed
- Our reporting
- High confidence
- Importance
- 2 of 5
- Last verified
- 26 October 2026
Your AI and this story
- GPT-6.2 Astra147 days after its cutoff
- Claude Opus 5.5117 days after its cutoff
- Gemini 4 Hydrogen208 days after its cutoff
- Grok 4.7147 days after its cutoff
None of these four assistants can know about it. The closest, Claude Opus 5.5, stops 117 days before it.
Key facts
- OpenAI: GPT-5.6 Luna $1/$6 → $0.20/$1.20 per M input/output tokens (Jul 30, −80%); GPT-5.6 Sol $5/$30 → $4/$20 for three months (Aug 21); GPT-6 Sol $2/$10 and GPT-6 Luna $0.10/$0.50 (Sept 22); GPT-6.1 Sol cache reads halved to $0.10 (Sept 29), about 75% cheaper cumulatively for the Sol line
- Anthropic: Sonnet 5 launch discount made permanent at $2/$10 (Aug 10); Fable 5.1 kept $10/$50 but cut cache reads 75% to $0.25 (Sept 1); Opus 5.5 at $4/$20 vs Opus 5’s $5/$25 (Sept 22)
- Argued causes: negligible switching costs through harnesses and OpenRouter, open-weight substitutes (the Luna cut seen as aimed at small open models such as DeepSeek V4 Flash), many hosts for open models (29 OpenRouter providers for DeepSeek V4.1 Flash 1017), tit-for-tat timing, and cheaper serving (Anthropic says Opus 5.5 needs less compute)
- Conclusion: prices will stay near marginal cost; subscription allowances are more likely to be cut than base API prices; both labs are seeking moats in enterprise commitments and personal-agent products
- Reach: 407 points on Hacker News (Oct 25); Techmeme headline Oct 26
What happened
The minimaxir.com blog collected the summer’s list-price changes for the two leading US labs in one table and asked why companies preparing IPOs would cut revenue per token. Its answer is that frontier APIs have become close to interchangeable. The prices are as tabulated in the post; individual figures are also in the linked release entries.
Why it matters
It is a compact record of the 2026 API price war, published the same week The Information reported OpenAI offering 30%+ discounts to win Anthropic customers such as Replit, and IBM’s CEO argued open-weight models are now good enough for most enterprise work.
Sources
2 sources from 2 sites. Numbers match the chips in the text.
2 sources: 2 reactions
Reactions
- minimaxir.com: OpenAI and Anthropic cut their LLM prices by up to 80% within a few months. Why?minimaxir.com, discussion
- Hacker News discussion (407 points)news.ycombinator.com, discussion
Changes
- Filed (Hacker News)