AitopkitOpenAI has shipped GPT-6 Sol and GPT-6 Luna — cheaper, faster versions of its GPT-6 Astra flagship — within hours of Anthropic's Claude Opus 5.5 launch. It is one of the most aggressive same-day responses the AI labs have ever staged, and it completes the GPT-6 family that began with Astra earlier this month.
The pattern is familiar from the GPT-5.6 era: a flagship model establishes the performance ceiling, then smaller siblings trade a slice of capability for speed and much lower cost per token, aimed at the everyday workloads that dominate real API bills. OpenAI claims the new models handle tasks substantially better than Anthropic's top offerings — a claim that, like Anthropic's, will be tested by independent benchmarks in the coming weeks.
The backdrop makes the timing interesting. The dueling launches arrived days after OpenAI CEO Sam Altman and Anthropic CEO Dario Amodei publicly debated whether AI development should slow down — and both companies then shipped cheaper, more capable models anyway. The frontier race is currently being fought as much on price-per-token as on raw intelligence, which is good news for anyone building on these APIs: re-benchmark your costs quarterly, because the "best value" slot keeps changing hands.
For developers choosing between the two new families: wait for third-party evals before migrating production workloads. Launch-day claims from both labs describe their own models at their best; the real comparison emerges once both are in the wild.