Claude Sonnet 4.5 retires November 30. Sonnet 5.5 takes over
New models on Complete
Anthropic retires Claude Sonnet 4.5 (claude-sonnet-4-5-20250929) on November 30, 2026, and names Claude Sonnet 5.5 (claude-sonnet-5-5) as its replacement. After that date, any request to the Sonnet 4.5 model ID fails. Claude Sonnet 5.5 is available for enterprise teams on Complete.
Claude Sonnet 4.5 | Claude Sonnet 5.5 | |
|---|---|---|
List price per million tokens in / out* | $3 / $15 | $2 / $10 |
Speed | 43 tokens per second | 141 tokens per second |
Independent score | 19 (estimate) | 56 at max effort |
Retirement | November 30, 2026 | Not before September 28, 2027 |
* Providers price per million tokens, which are fragments of words.
Scores and speeds come from Artificial Analysis. Its score for Sonnet 4.5 is an estimate, and it does not publish a cost per task for Sonnet 4.5.
What the switch changes
List price drops by a third, from $3 in and $15 out to $2 in and $10 out.
Anthropic cut the price of cached input to $0.10 per million tokens on October 7, and says that makes most agent work about 20% cheaper.
Artificial Analysis measures Sonnet 5.5 at 141 tokens per second on Anthropic's API, against 43 for Sonnet 4.5. That is over 3x faster.
It reads text and images and holds up to 1 million tokens in a single request.
Artificial Analysis lists 9 API providers for it, including Amazon Bedrock and Google Cloud Vertex AI.
Amazon Bedrock and Google Cloud set their own retirement dates for Sonnet 4.5. Anthropic's November 30 date applies to the Claude API, Claude Platform on AWS and Microsoft Foundry.
Effort decides the cost
Sonnet 5.5 scores 56 on the Artificial Analysis Intelligence Index at max effort, second of 226 models. The index is an independent rating built from 10 tests covering agent work, coding, general knowledge and science.
On the Artificial Analysis leaderboard, cost per task rises about 18x from the lowest setting to the highest:
Effort setting | Score | Cost per task |
|---|---|---|
Max | 56 | $7.67 |
Extra high | 52 | $2.75 |
High | 47 | $1.12 |
Medium | 41 | $0.59 |
Low | 36 | $0.42 |
Extra high scores 52 for $2.75, close to GPT-6.1 Sol (52 for $0.72) and Gemini 4 Argon (53 for $1.99). Max effort is where the last points come from, and where most of the cost sits.
Some steps should move to Haiku instead
Not every Sonnet 4.5 workflow needs Sonnet 5.5. Claude Haiku 5.5, released October 7, scores 43 at max effort for $0.21 per task. That beats Sonnet 5.5 on its medium setting (41 for $0.59) at about a third of the cost. High volume steps such as classification, extraction and routing are candidates to move down a tier.
A migration plan before November 30
Find every workflow pinned to claude-sonnet-4-5-20250929.
Pick an effort setting per workflow. Start at medium or high and move up only where results fall short.
Move simple, high volume steps to Claude Haiku 5.5.
Remove temperature, top_p and top_k from requests. Anthropic's newer models return an error when they are set.
Switch before the cutoff, with time to compare results on real work.
Run it on your own workflows
Claude Sonnet 5.5 is available for enterprise teams on Complete. Book a demo to move your Sonnet 4.5 workflows before November 30.
