Overview
- Both vendors launched the new models on Sept. 22, with Anthropic rolling out Claude Opus 5.5 and OpenAI debuting GPT‑6 Sol and GPT‑6 Luna.
- Publishers set materially lower token prices: OpenAI priced Sol at $2/$10 per million input/output tokens and Luna at $0.10/$0.50, while Anthropic listed Opus 5.5 at $4/$20 per million tokens.
- Each company published vendor-run benchmarks claiming efficiency gains and lower cost‑per‑task, with Text Arena and other community leaderboards showing Opus 5.5 early favorability and OpenAI highlighting Sol’s improved agent completion rates.
- The releases change practical engineering choices by lowering the cost of long context windows, caching and repeated agent calls, which could let teams run longer agents and larger code migrations for far less money.
- Coverage stresses that most performance and cost claims are vendor‑reported so enterprises should validate on their own workloads, update caching and orchestration, and add monitoring and safety controls before scaling.