New Delhi: Anthropic has released a new AI model called Claude Opus 5. The company says Opus 5 brings notable gains in efficiency and cost-effectiveness compared to its earlier model, Opus 4.8, while keeping the same pricing.
The model was launched recently and is priced at $5 per million input tokens and $25 per million output tokens, the same as Opus 4.8. In comparison, Fable 5 costs $10 per million input tokens and $50 per million output tokens.
Anthropic describes Opus 5 as a model built for everyday use, one that works more efficiently than other models. It is now the default model on Claude Max and the strongest model available on Claude Pro.
On performance, Anthropic said Opus 5 sets a new high mark among its own models for coding and knowledge work, though it still trails Mythos 5 on cybersecurity tasks. On one test called Frontier-Bench v0.1, the company said Opus 5 beats every other model and more than doubles the score of Opus 4.8, all at a lower cost per task.
On another test, CursorBench 3.2, it comes within half a percent of Fable 5’s best score, at half the price. On Frontier Bench v0.1 specifically, Opus 5 scored 43.3 percent compared to Fable 5’s 33.7 percent.
Anthropic compared Fable 5 and Opus 5 across 13 tests. It said Opus 5 scored higher on eight of them, even though it costs around half as much.
On safety, Opus 5’s cyber-related restrictions are less strict than those placed on Fable 5. It can find flaws in source code but is blocked from binary-based vulnerability scanning, penetration testing, and creating exploits. Anthropic expects these safety checks to step in about 85 percent less often than they do with Fable 5. Like earlier Opus models, Opus 5 has no data retention requirement for general users, something the company highlighted for customers who need a strict zero data retention policy.
On why this launch matters, VentureBeat noted that Anthropic is not calling Opus 5 its smartest model, since that title still belongs to Fable 5. Instead, the company argues that most valuable AI work falls in a middle range of difficulty, where near-top intelligence delivered cheaply and quickly matters more.
The launch adds to the ongoing competition between Anthropic, OpenAI, and China’s Moonshot, as AI companies push toward cheaper, near-frontier models. An Anthropic product leader also told Reuters that the release shows the company’s fast pace of development.
Previous
