Claude Opus 5 nears Fable 5 performance at half the price
By Steven Van ·
Pricing stays at $5 per million input tokens and $25 per million output, unchanged from Opus 4.8, and it's now the default model on Claude Max.
Anthropic has released Claude Opus 5, and the pitch is unusually direct: intelligence close to its flagship Claude Fable 5 at roughly half the cost. Opus 5 is now the default model on Claude Max, the strongest model available on Claude Pro, and live everywhere Claude runs, including the API, Claude Code, and Claude Cowork.
Doubling Opus 4.8, closing in on Fable
The benchmark deltas are large for a same-price successor. On Frontier-Bench v0.1, Anthropic's hardest agentic evaluation, Opus 5 doubles Claude Opus 4.8's score while costing less to run per task. On CursorBench 3.2 it lands within 0.5% of Fable 5 at half the price, and on OSWorld 2.0, the computer-use benchmark, it outperforms every model at its cost level and even surpasses Fable 5 while spending a third as much. Its ARC-AGI 3 score comes in three times higher than competing models. On coding and knowledge-work evals like Frontier-Bench and GDPval-AA, Axios notes Opus 5 is the new state of the art, with only the restricted Mythos 5 ahead of it on cybersecurity tasks.
The judgment upgrade
Beyond scores, Anthropic and early customers keep returning to one word: judgment. Opus 5 is described as a thoughtful, proactive model that catches logical errors during planning rather than after execution, verifies its own work, and stays consistent across long, multi-step agentic tasks. It also brings notable gains in scientific domains, including life sciences, organic chemistry, and protein analysis, and produces the most sophisticated visual output of any Opus so far, from data visualizations to polished front-end work.
Pricing, Fast mode, and availability
Pricing is unchanged from Opus 4.8: $5 per million input tokens and $25 per million output. That is the entire strategy in one line, a big capability jump at last generation's price, which TechCrunch reports makes Opus 5 the rational pick over Fable 5 for most workloads. Opus 5 also supports Fast mode, which serves the same model at 2.5x output speed for 2x the base price. The ecosystem moved on day one: Cursor, v0, Lovable, Notion, Perplexity, Rork, and Base44 all shipped Opus 5 support within hours of the announcement.
Four models in sixty days
Opus 5 is Anthropic's fourth model release in under two months, following Sonnet 5, Fable 5, and Mythos 5 in June. Each targets a distinct tier rather than replacing the last: Mythos and Fable at the frontier, Opus as the near-frontier workhorse, Sonnet as the agentic mid-tier. For creators and builders, the practical takeaway is that the model doing most of the work inside your tools just got significantly smarter without your bill changing.
Sources: Anthropic, TechCrunch, Axios.