Anthropic · Claude Opus
Claude Opus 5
Anthropic’s current model for complex agentic coding and enterprise work, and the model it points Opus 4.8 users at. Thinking is on by default, so the effort setting, not a thinking switch, is how you control depth and cost.
This is the current Opus, so a migration is a real decision rather than a version bump. Test the two breaking changes first — thinking on by default and disabling thinking rejected above high effort — then re-baseline cost, because Anthropic itself says responses run longer.
Use it for
- Complex agentic coding and long-horizon tasks
- Deep multi-step reasoning and code review
- Enterprise work with heavy context and documents
Do not treat it as a drop-in for Opus 4.8: thinking is now on unless you disable it, which changes what fits inside max_tokens. It is also poor value where Sonnet 5 already passes the task.
Effort without the jargon
The work is routine enough that a shallower pass holds quality at a fraction of the tokens and latency.
This is the default on the Claude API and Claude Code, and the sensible starting point before tuning in either direction.
The task is genuinely demanding and you can give it a large max_tokens; note that thinking cannot be disabled at these levels.
Advanced recordCost, limits, privacy and operations
Opus 5 costs the same per token as Opus 4.8 but thinks by default, so the same request can produce more billable tokens; measure total spend on your own work rather than reading the unchanged rate card as an unchanged bill.
- API cost
- Officially disclosedClaude API list price: US$5 per million input tokens and US$25 per million output tokens, unchanged from Opus 4.8. Fast mode is a research preview on the Claude API only, priced at US$10 and US$50; caching, batch and cloud-platform rates differ.
- Context window
- Officially disclosed1,000,000 tokens (both the default and the maximum; there is no smaller context variant)
- Maximum output
- Officially disclosed128,000 tokens (synchronous Messages API)
- Knowledge cutoff
- Officially disclosedReliable knowledge: May 2026; training data: May 2026
- Recorded inputs
- Text · Images
- Recorded output
- Text
Data handling is a product decision
Depends on surfaceAnthropic says commercial API inputs and outputs are deleted from its backend within 30 days by default, with exceptions for stored services, agreed controls, safety enforcement and law. Consumer Claude and cloud platforms have separate terms.
Operational constraints
- Thinking is on unless disabled, and max_tokens is a hard limit on thinking plus response text, so budgets sized for Opus 4.8 need revisiting.
- Disabling thinking is rejected with a 400 error above high effort, checked on every request.
- Effort defaults to high on the Claude API and Claude Code; the full ladder is low, medium, high, xhigh and max.
- The minimum cacheable prompt is 512 tokens, down from 1,024 on Opus 4.8.
- Anthropic says Opus 5 verifies its own work, so verification instructions carried over from earlier models cause over-verification.
Unknown means the cited official record does not disclose a safe value. It is not an estimate. Prices are provider list prices in USD where stated and can change before this record's review date.
Keep the evidence separate
Anthropic describes Opus 5 as a step-change over Opus 4.8, with the largest gains in deep reasoning, agentic and long-horizon tasks and test-time compute scaling, at US$5 and US$25 per million tokens — unchanged from Opus 4.8 and half the cost of Fable 5. It has a one-million-token context window that is both default and maximum, 128k max output, and thinking on by default.
No sufficiently useful independent note has been added yet. That absence is not filled with our guess.
No Human Bit result is claimed; use this as evaluation guidance only.
Not scheduled
No Human Bit result is claimed for this model yet.