Claude Sonnet 4.5 — reviews, specs & pricing
Anthropic's strongest all-purpose Sonnet: excellent reasoning, coding, and long-context work at a mid-tier price.
Summary
Claude 4.5 Sonnet is Anthropic's premier mid-tier model designed to deliver elite-level reasoning, coding, and comprehension capabilities. It excels at deeply understanding complex prompts, maintaining context over large datasets, and producing highly articulate responses. This makes it an ideal balance of state-of-the-art intelligence and cost efficiency for demanding enterprise workflows.
Sample use case
A software development team can employ Claude 4.5 Sonnet to automate the refactoring of a legacy codebase by feeding it entire directories of old code. The model can accurately map dependencies, identify security vulnerabilities, and generate clean, updated code alongside comprehensive unit tests, dramatically accelerating the modernization process.
Specifications
- Provider: anthropic
- License: closed
- Context: 200k tokens
- Input price: $3/M tok
- Output price: $15/M tok
Pros
- Exceptional logic and coding proficiency
- Large, highly accurate context processing
- Superior instruction-following and nuance
- Distinguished balance of performance and price
Cons
- Slower speed compared to lightweight models
- Lacks built-in media generation features
- Premium pricing relative to basic utility models
Average rating 3.4 from 5 community reviews on Reviuws.
Community reviews
Simon Willison's blog on Claude Sonnet 4.5
Rating: 4.0 / 5 — by Simon Willison (use case: software and agentic coding)
With preview access before release, Willison calls Sonnet 4.5 'probably the best coding model in the world (at least for now)', noting it felt only incrementally better than Sonnet 4 but performed strongly across his usual coding evaluations.
Pros: Consistently strong on his own coding benchmarks; real improvements.
Cons: Gains feel incremental rather than transformative; claim is time-limited.
TechCrunch on Claude Sonnet 4.5
Rating: 3.0 / 5 — by Maxwell Zeff (use case: production application development)
TechCrunch reports Sonnet 4.5 as Anthropic's best coding model, said to build production-ready applications rather than prototypes, launched alongside Claude Code 2.0.
Pros: Claimed production-ready app building; paired with improved Claude Code tooling.
Cons: Claims sourced from Anthropic; no hands-on testing.
Ars Technica on Claude Sonnet 4.5
Rating: 3.0 / 5 — by Ars Technica staff (use case: long-horizon agentic tasks)
Ars covers Anthropic's claim that Sonnet 4.5 held focus for 30 hours on multistep agentic tasks, with improved coding and computer use, treating the figure as notable but company-sourced.
Pros: Sustained long multistep agentic work; better computer use.
Cons: 30-hour claim not independently reproduced; little on failure modes.
Tom's Guide on Claude Sonnet 4.5
Rating: 4.0 / 5 — by Amanda Caswell (use case: autonomous coding agents)
Tom's Guide frames the 30-hour autonomous coding claim as potentially transformative for how work gets done, with enthusiasm tempered by the fact it is an early, company-driven claim.
Pros: Potential for long autonomous coding sessions; strong enterprise agent positioning.
Cons: Announcement-driven; long-run reliability unverified.
PCMag on Claude Sonnet 4.5
Rating: 3.0 / 5 — by PCMag news desk (use case: enterprise coding with safety focus)
PCMag highlights that beyond coding gains, Anthropic touted reduced deceptive behaviour in Sonnet 4.5, framing the release as an improvement in both capability and honesty.
Pros: Reported reduction in deceptive or sycophantic behaviour; strong coding claims.
Cons: Honesty claim hard to verify from a news brief; largely a rehash of the announcement.
Frequently asked questions
What is Claude Sonnet 4.5?
Claude 4.5 Sonnet is Anthropic's premier mid-tier model designed to deliver elite-level reasoning, coding, and comprehension capabilities. It excels at deeply understanding complex prompts, maintaining context over large datasets, and producing highly articulate responses. This makes it an ideal balance of state-of-the-art intelligence and cost efficiency for demanding enterprise workflows.
How much does Claude Sonnet 4.5 cost?
Claude Sonnet 4.5 costs $3 per million input tokens and $15 per million output tokens.
Is Claude Sonnet 4.5 open source?
Claude Sonnet 4.5 is released under the closed license.