Tools · Model Watch
Opus 5.5 lowers the API cost of another coding attempt
Anthropic’s new model arrives with lower token prices and stronger coding claims. For creators, the useful test is what survives the next revision.
Reporting updated Sep 23, 2026
THE TAKEAWAY
Lower token prices make retries cheaper at the same token usage. They do not tell you how much work a generated scene will need.
What Changed
Anthropic released Claude Opus 5.5 on September 22 at $4 per million input tokens and $20 per million output tokens. Both prices are 20% below Opus 5. The company also reports stronger graphics and polish in an early tester’s game-building experiment.
Those are different kinds of evidence. The API prices tell you what tokens cost. The demonstration gives you a reason to try the model; it does not establish what your finished project will cost.
Why Creators Should Care
If you use the same number of tokens, a retry costs less. That gives a small team more room in the budget to revise a game interaction or rebuild a scene. But a cheaper attempt is only useful if it gets the project closer to something you can keep.
VEYR would test that with an existing project. Ask for one bounded change, such as adding an adjustable light without moving the camera. Then inspect the result and request a revision. A model that can preserve the parts you already like is more useful than one that keeps replacing the whole scene.
What Remains Unproven
The release does not establish which model is best for your 3D workflow. A generated scene can look convincing and still need repairs before another creator can work on it. More attempts, longer answers or extra cleanup could erase a token-price saving.
Our next questions are practical: does the project still open, are its controls understandable, and can another creator edit the result? We have not run that comparison for this report. Treat launch demonstrations as reasons to test, not evidence that cleanup has disappeared.
.webp)