Quick answer: On September 22, 2026, Anthropic launched Claude Opus 5.5 at roughly 40% lower cost than Opus 5 while matching Fable 5.1-level performance, and OpenAI released GPT-6 Sol and GPT-6 Luna with API pricing 50% below current GPT-5.6 promotional rates. The two releases landed within hours of each other, intensifying a price war just days after both companies’ leaders publicly called for slowing frontier AI development.
What happened
Anthropic’s Opus 5.5 launched first, with Sonnet 5.5 and Haiku 5.5 slated for coming weeks. OpenAI followed with GPT-6 Sol and GPT-6 Luna, variants of GPT-6 Astra, undercutting Anthropic’s new pricing. Both companies framed the cuts as passing efficiency gains directly to users rather than reducing model capability.
The timing is notable: these releases come shortly after Anthropic CEO Dario Amodei’s essay calling for the industry to slow frontier development, and the same week OpenAI, Anthropic, and Google DeepMind confirmed talks about a joint AI safety standards body. Competing on price for incremental, cheaper models is a very different kind of race than the frontier-capability race the safety talks were meant to cool down.
Why it matters
For any business or SEO team running AI-assisted workflows, this is a direct, near-term cost story: API pricing for comparable capability is dropping fast on both major platforms simultaneously. An economist quoted this week noted the dynamic could compress both companies’ ability to profit even as usage grows, meaning further price moves are likely.
Key facts
- Anthropic: Claude Opus 5.5, about 40% cheaper to run than Opus 5
- OpenAI: GPT-6 Sol and GPT-6 Luna, API pricing 50% below GPT-5.6 promo rates
- Both released within hours of each other on September 22, 2026
- Comes days after both companies’ leaders called for slowing frontier AI development
FAQ
Are these frontier model breakthroughs?
No — both companies describe these as incremental, efficiency-focused releases rather than major capability leaps.
Does this contradict the AI slowdown talk?
It’s a notable contrast: leadership has publicly called for slowing frontier development while simultaneously competing hard on price and shipping new models within hours of each other.
Related reading
- OpenAI, Anthropic and Google DeepMind Move Toward an AI Safety Standards Body
- Antitrust Suit Accuses OpenAI, Anthropic, SpaceXAI and Google of an Illegal AI Slowdown Pact
- More AI industry coverage
Source: Fortune
