Claude Opus 5.5 beats OpenAI's Astra on coding and price

Anthropic has launched Claude Opus 5.5, and according to the published tests, the model performs better than OpenAI's new GPT-6 Astra on coding. It is also cheaper to use.

What happened?

OpenAI released GPT-6 Astra on September 3, 2026. Nineteen days later, Anthropic responded with Claude Opus 5.5, the first model in the Claude 5.5 generation. At the same time, Anthropic has cut the price by about 40 percent compared with its predecessor Opus 5.

The numbers

A benchmark is a standardized test that makes it possible to compare AI models. On Terminal-Bench 4.0, which measures how well an AI can solve tasks in a command line (the text-based way of controlling a computer), Opus 5.5 scored 66.4 percent. Astra scored about 58 percent.

On FrontierCode v1.1, which tests coding across entire software projects, the difference is small: 54.4 percent for Opus 5.5 versus 53.3 percent for Astra. On a test of ordinary knowledge work, GDPval-AA, Opus 5.5 scored 1,846 points versus Astra's 1,542.

The price

AI models are typically billed per million tokens, small chunks of text. Opus 5.5 costs $4 for input and $20 for output. Astra costs $10 for input and $50 for output. That makes Opus 5.5 about 2.5 times cheaper.

Anthropic itself says that Opus 5.5 at its standard setting matches Astra's performance at about 40 percent of the price.

What does it mean for you?

If your workplace uses AI for programming, the price difference can be noticeable, because AI coding often uses large amounts of tokens. For ordinary chatbot users, it mostly means that competition is pushing prices down.

The figures should be taken with a grain of salt, however. Many of them come from the makers' own tests, and OpenAI has not published results on all the same measures. According to a comparison from Vellum and others, Astra is strong in the areas it is tuned for, so the best model depends on the specific task.

What this means for you

If you or your colleagues use AI to write code, this is where you will feel the difference, because that kind of work eats a lot of tokens, and Opus 5.5 costs about 2.5 times less than Astra. Ask your IT department or vendor whether they have compared the two models on your own tasks before you choose. If you mostly use a chatbot for emails and texts, the gain is less direct, but sharper competition between Anthropic and OpenAI typically pulls prices in your favor.