Anthropic and OpenAI released new AI models this week, and both companies are cutting prices along with the upgrades. Claude Opus 5.5 replaces Anthropic’s Fable 5.1 as the company’s flagship model, while OpenAI’s GPT-6 Sol and GPT-6 Luna follow its GPT-6 Astra release from earlier this year.
The releases follow months of industry debate over whether AI development should slow down. Instead, Anthropic and OpenAI are shipping models that match or beat their predecessors on benchmarks while charging less per token, a sign that competition between the two companies keeps intensifying on price as much as raw capability.
What claude opus 5.5 does better
Anthropic says Opus 5.5 handles complex work more reliably, including finding and fixing inefficiencies in software, financial analysis and general business tasks. The company’s own benchmarks show Opus 5.5 outperforming GPT-6 Astra on agentic coding tests, including Terminal-Bench 4.0 and FrontierCode v1.1, results that could sway developers weighing which model to build with.
Anthropic also points to changes beyond raw task performance. The company says Opus 5.5 writes more clearly than earlier models and that it “attempted to circumvent boundaries around 85 percent less often” than its predecessors. Pricing fell too: Anthropic says Opus 5.5 costs $4 for input tokens and $20 for output tokens, down from $5 and $25 with Opus 5.
Openai’s gpt-6 sol and gpt-6 luna
OpenAI’s new models chase the same tradeoff. GPT-6 Sol and GPT-6 Luna were trained using methods similar to GPT-6 Astra and improve on professional work, coding and computer use, but cost up to 50 percent less than the promotional pricing OpenAI offered for its GPT-5.6 models. GPT-6 Sol costs $2 for input tokens and $10 for output tokens, while GPT-6 Luna costs $0.10 for input tokens and $0.50 for output tokens.
Both models are more accurate than their predecessors, according to OpenAI. GPT-6 Sol makes about half as many factual errors as the model it replaces, and OpenAI says it can match Fable 5.1’s coding performance at a lower cost. Faster response times and expanded prompt caching let the models reuse more context between requests.
Availability and safety claims
OpenAI also says the GPT-6 models are more aligned with company guidelines than earlier versions, based on its internal testing. The models lie less often about the results of coding work and refuse more attempts to bypass safety limits, according to OpenAI. Like Anthropic, the company has proposed its own criteria for third-party evaluators to use when judging AI safety and progress.
Opus 5.5 is available now through Claude, Amazon Web Services, Google Cloud and Microsoft Azure. GPT-6 Sol and GPT-6 Luna are live in ChatGPT Work and Codex for Plus, Pro, Business and Enterprise subscribers. Free ChatGPT users and Go subscribers get access only to GPT-6 Luna through OpenAI’s desktop app.