Anthropic and OpenAI Unleash Rival Frontier Models as AI Leaders Tout Capability Pacing

Anthropic and OpenAI have released competing high-performance AI models within weeks, even as both companies publicly advocate for deliberately slowing the pace of AI capability advancement. Anthropic launched Claude Opus 5.5 on September 22, 2026, claiming it matches the intelligence of its larger Fable 5.1 model while cutting costs by 40 percent. OpenAI countered with a three-tier response: GPT-6 Astra (released September 3), optimized for complex reasoning and computer control, followed by GPT-6 Sol and GPT-6 Luna (both September 22), driving pricing 50 percent lower across the board.