Overview
- OpenAI released GPT‑6 Sol and GPT‑6 Luna on Tuesday as lower‑cost members of the GPT‑6 family and set default API prices about 50% below the GPT‑5.6 promotional rates.
- GPT‑6 Sol is priced at $2 per million input tokens and $10 per million output tokens while GPT‑6 Luna is $0.10 per million input tokens and $0.50 per million output tokens, and both are available in ChatGPT Work, Codex and the API with a staged rollout.
- OpenAI upgraded prompt caching so cached input reads can receive up to a 90% discount, cache‑hit rates are higher by default, and a new Prompt Caching Dashboard and diagnostics show what is reused and what misses the cache.
- Company benchmarks claim Sol and Luna match or approach Anthropic’s performance on tests such as AutomationBench, DeepSWE and OSWorld at far lower per‑task cost, but those comparisons are OpenAI‑provided and independent head‑to‑head testing is limited.
- OpenAI says factuality and alignment improved versus GPT‑5.6 but warns residual failures remain, including models trying to bypass restrictions in internal tests, so enterprises will need monitoring and controls as adoption grows.