source: Simon Willison: Claude Opus 5.5, GPT-6 Sol, GPT-6 Luna, and a new price war
level: technical
anthropic released claude opus 5.5, and openai followed with gpt-6 sol and gpt-6 luna. the new models bring major price cuts. gpt-6 luna costs $0.10 per million input tokens and $0.50 per million output tokens, half the price of gpt-5.6 luna. gpt-6 sol is also half the price of gpt-5.6 sol. claude opus 5.5 dropped 20 percent to $4 per million input and $20 per million output, with cache reads falling 60 percent.
the pricing table shows gpt-6 luna is one of openai's cheapest models ever, beaten only by gpt-4.1 nano and gpt-5 nano. grok 4.7, priced at $2 input and $6 output, now matches gpt-6 sol on input. claude opus 5.5 at max thinking level failed to return a response on a simple svg test, hitting its 128,000 token output limit twice. each failure cost $2.56 and took nearly 20 minutes, suggesting max mode may be unreliable for complex tasks.
the price war affects the tier below top models like gpt-6 astra and claude fable 5.1, both at $10 input and $50 output. anthropic plans to release sonnet 5.5 and haiku 5.5 soon, but haiku 4.5 at $1 input and $5 output is ten times more expensive than gpt-6 luna. simon willison now uses gpt-6 sol and claude opus 5.5 as defaults in codex and claude code, and upgraded the datasette agent demo to gpt-6 luna, finding it fast and competent.
why it matters: lower prices make advanced ai models more accessible for building applications, but reliability issues like over-thinking can limit their usefulness.
source: Simon Willison: Claude Opus 5.5, GPT-6 Sol, GPT-6 Luna, and a new price war