claude-opus-4-5 — pricing, context window and limits
A legacy Opus that still answers: the first model with an effort parameter, but with only three levels — xhigh and max both error here.
Price per million tokens
| Input | $5 |
|---|---|
| Output | $25 |
| Cache write, 5-minute TTL | $6.25 |
| Cache write, 1-hour TTL | $10 |
| Cache read | $0.50 (0.1x input) |
| Batch input / output | $2.50 / $12.50 |
Limits and behaviour
| Status | Legacy — still answers |
|---|---|
| Context window | 200K tokens |
| Max output | 64K tokens |
| Min cacheable prefix | 4,096 tokens |
| Effort levels | low, medium, high |
| Effort default | high |
| Thinking | Fixed budget_tokens, off unless requested |
| Forced tool_choice | Accepted |
| Platforms | Claude API |
Lifecycle
| API model name | claude-opus-4-5-20251101 |
|---|---|
| Retires no sooner than | 24 Nov 2026 |
| Successor | Claude Opus 5.5 |
Still active, and the first Opus with effort control at all
Opus 4.5 introduced the effort parameter to the Opus line, but only at three levels — the top
tier and the one below it both error if requested. It remains an active, legacy model rather than a
deprecated one, distinct from the older generations further back that have now been retired on the
Claude API.
How far behind the current Opus it sits
Compared with Opus 5.5, this generation carries the older, fixed-budget thinking, a narrower effort range, a smaller context window and a much higher minimum cacheable prefix — at a per-token price above Opus 5.5's. There is no longer a price argument for staying: the current Opus is cheaper per token and caches shorter prompts.
Why teams on this generation tend to stay anyway
Because it is still active, nothing forces a move the way a retirement does, and "it still works" quietly substitutes for a decision. The lifecycle table above shows the earliest date this model can be retired; a migration planned before that floor passes is cheaper than one compressed into a notice period. The migration guide names Opus 5.5 as the target, reached by applying each Opus generation's changes in order.
Related
See what actually changed from Opus 4.6 to Opus 5 for the largest behavioural jump since this generation, and Claude model retirement dates for where this model sits in the calendar.
Verified 2026-09-30 against claude-api skill — shared/models.md, model-migration.md + SKILL.md model table and Anthropic's pricing and model-deprecation pages (canonical: https://platform.claude.com/docs/en/about-claude/models/overview, https://platform.claude.com/docs/en/about-claude/pricing, https://platform.claude.com/docs/en/about-claude/model-deprecations).Could not confirm: Context window and max output for this legacy model are not restated in our source; the figures shown are the documented defaults for its generation and should be re-checked against the Models API before you budget against them.
Verified 2026-10-28 against Anthropic's model-deprecations status table and pricing page (canonical: https://platform.claude.com/docs/en/about-claude/model-deprecations, https://platform.claude.com/docs/en/about-claude/pricing).Could not confirm: These dates apply to the Claude API, Claude Platform on AWS, Claude Models on Google Cloud (Anthropic managed) and Microsoft Foundry. Amazon Bedrock and Claude Models on Google Cloud (Google managed) keep their own retirement schedules, which we do not track.