AI GlossaryㅇWords you meet while using AI
Opus 4.8
A version of an AI model from Anthropic, a top-tier model used for coding and agentic tasks.
In plain words
Opus 4.8 is a version of an AI model made by Anthropic. Even from the same company, a different number roughly means "a later, more experienced version," the way a person gains experience over time. 4.8 is one point in the higher-tier lineup among the company's various models.
When Vercel, an intermediary service, tallied AI model usage across companies, Opus 4.8 ranked #1 in spending, ahead of even Anthropic's pricier model, Fable 5. That means it's a model people actually pay for and use heavily.
In another experiment, researchers ran multiple instances of Opus 4.8 at once and had them collaborate with each other. The results were interesting. Earlier versions often ran into conflicts when several instances touched the same piece of work, causing output to be discarded. Opus 4.8 rarely had this problem. But it turned out the reason wasn't that it collaborated well — it was that it avoided collaborating in the first place so as not to collide with the others.
How it shows up in the news
The article states: "In Vercel's July gateway data, Fable 5 ranked second behind Opus 4.8." It's easy to assume Opus 4.8 is the most expensive model in the company's lineup, but Fable 5 and Opus 5, which also appear in the same article, are separate models released later than, or priced higher than, Opus 4.8. A higher version number doesn't automatically mean newer or better.
See also
Stories using this term
- Anthropic launches developer hub claude.devAI · 2026.10.01
- Anthropic publishes Sonnet 5.5 developer guideAI · 2026.09.30
- Anthropic releases results of agent trading experimentAI · 2026.09.25
- Anthropic Launches Claude Opus 5.5AI · 2026.09.23
- OpenAI beats Alibaba on Alibaba's own benchmarkAI · 2026.09.04
- Anthropic has Claude tackle AI alignment research, and it outperforms humansAI · 2026.08.31
