OpenAI and Anthropic cut AI model prices as competition intensifies

  • OpenAI launched GPT-6 Sol and Luna after Anthropic released Claude Opus 5.5.
  • GPT-6 Sol costs $2 input and $10 output per million tokens.
  • Anthropic cut Opus 5.5 cache-read pricing 60% from Opus 5.

OpenAI introduced GPT-6 Sol and GPT-6 Luna, cutting their API prices by up to 50% from GPT-5.6 promotional rates, hours after Anthropic launched Claude Opus 5.5 and said it costs 40% less to run than its predecessor. The back-to-back releases highlight an industry shift toward cost-per-task and efficiency as companies deploy AI agents for coding, research, computer use and other extended workflows. Sol costs $2 per million input tokens and $10 per million output tokens, while Luna costs $0.10 and $0.50, respectively. OpenAI positioned GPT-6 Astra as its highest-performance model and said Sol scored 33.2% on AutomationBench at a reported cost of $0.27 per task, compared with Astra's 30.3% score at low effort and 3.9 times the cost. Anthropic said cache reads represent most costs in agentic and coding workloads and cut its cache-read price by 60% from Opus 5. Earlier information in the topic record said Opus 5.5 was released at 16:31 UTC, followed by OpenAI's models at 18:12 UTC, 101 minutes later, and that both launches came 10 days after Dario Amodei advocated slower capability gains, a position endorsed by Sam Altman and supported by Elon Musk.

The information on this website is generated using AI and we cannot guarantee its accuracy. Please use it as reference information only.