ClaudeHowSupport Us

claude-sonnet-4-5 — pricing, context window and limits

Statuslegacy
Input / MTok$3
Output / MTok$15
Context window200K tokens
Max output64K tokens
Min cacheable prefix1,024 tokens
Effort levelsnot supported
Thinkingbudget-tokens
PlatformsClaude API, Amazon Bedrock, Google Cloud

Active, but with no effort control at all

Sonnet 4.5 predates the effort parameter's arrival on the Sonnet line entirely — there's no tiered reasoning control here, only the older thinking-budget mechanism. A migration from this model to Sonnet 4.6 inherits that generation's default effort level automatically, which can quietly raise both latency and spend if nobody deliberately reviews the new default against what this model's absence of the parameter previously meant in practice.

Why this one is easy to under-plan a migration from

Because this model has no effort parameter to migrate away from, it's tempting to treat a move off it as simpler than moving off a model that does have one — in practice it's the opposite: there's no existing setting to carry forward as a starting point, which means the destination model's default is being adopted blind rather than deliberately chosen.

Confirm the current defaults rather than carrying assumptions forward

Because this model's behaviour predates so much of what the current lineup treats as standard, don't assume anything about default thinking behaviour, cache minimums, or effort support carries over from working with newer models — check this page's own table directly before writing new code against it.

See migrating: Sonnet 4.5 to Sonnet 5 for the direct diff to the current generation.

Verified 2026-08-08 against claude-api skill — shared/models.md + SKILL.md model table (canonical: https://platform.claude.com/docs/en/about-claude/models/overview, https://platform.claude.com/docs/en/pricing).Could not confirm: Context window and max output are not restated in our source for this model; re-check via the Models API before budgeting.