Claude Opus 5
2026-07-24 · Publisher : Anthropic · Domain : Artificial Intelligence
Anthropic launches Claude Opus 5, a thoughtful and proactive model close to the intelligence of Claude Fable 5 at half the price, at the same pricing as Opus 4.8.
What changes
- State-of-the-art performance announced in coding and knowledge work, with a better cost/efficiency ratio than Opus 4.8.
- Checks its work and iterates until it succeeds; stronger visual outputs.
- Presented as Anthropic's most aligned model, with the lowest rates of deceptive behavior.
- Fast mode at twice the base price, about 2.5 times faster.
### Specifications - Context window: 1M tokens. Maximum output: 128,000 tokens. Thinking enabled by default. - Knowledge cutoff: not disclosed for this launch. - Availability: claude.ai, API, Claude Code, Cowork, Bedrock, Google Cloud, Foundry. API identifier: `claude-opus-5`. ### Pricing | Item | Pricing (USD per million tokens) | | --- | --- | | Input | 5 | | Output | 25 | | Fast mode input / output | 10 / 50 | | Cache write 5 min / 1 h | 6.25 / 10 | | Cache read | 0.50 | | Batch (input / output) | 2.50 / 12.50 | ### Benchmarks Scores published by the vendor, mostly expressed in relative terms in the post: | Evaluation | Result | | --- | --- | | Frontier-Bench v0.1 | best score, more than 2 times Opus 4.8 | | CursorBench 3.2 | within 0.5% of Fable 5 at half the cost | | ARC-AGI 3 | 3 times the next best model | | OSWorld 2.0 | surpasses Fable 5 at a third of the cost | | Organic chemistry (internal) | +10.2 points vs Opus 4.8 | ### Key takeaways Opus 5 delivers most of Fable 5's performance at Opus pricing, making it the most cost-effective agentic option before Opus 5.5.
Links
Tags : LLM, Anthropic, Claude, Opus, Agents, Codage