Anthropic releases Claude Sonnet 5.5, claiming over 30% faster performance
Anthropic has released Claude Sonnet 5.5 as an update to its mid-range Claude model. The company says it is over 30% faster than Sonnet 5 while keeping the same prices, starting at $2 per million input tokens.
Image: itc.ua · Author: \u0428\u0430\u0434\u0440\u0456\u043d \u0410\u043d\u0434\u0440\u0456\u0439 · source articleEditorial excerpt for news reporting
Anthropic has released Claude Sonnet 5.5, replacing the Sonnet 5 model introduced in June. The company positions the new release as its mid-range model for everyday tasks with clearly defined boundaries.
Anthropic says Sonnet 5.5 produces results more than 30% faster than its predecessor and typically uses fewer tokens for the same work. The company calls it the fastest Sonnet model to date.
The model is intended for tasks including bug fixing, document, presentation and spreadsheet creation. Anthropic also highlights interface improvements and template-based presentation generation.
Sonnet 5 pricing remains unchanged
The prices are $2 per million input tokens, $10 per million output tokens and $0.20 per million cache-read tokens.
On Terminal-Bench 4.0, Sonnet 5.5 exceeded Sonnet 5’s best result while using less than one-tenth of the cost for the task. On FrontierCode, Anthropic says it reaches GPT-6 Sol’s result at about one-fifth of the cost of a single task.
Results in workplace benchmarks
In GDPval-AA, which covers tasks from 44 professions across nine industries, Sonnet 5.5 scored nearly as well as Opus 5.5 and about 400 points ahead of Sonnet 5.
In a separate internal test, the model prepared a 10-slide presentation using company financial materials, a conference-call transcript and a finished template. Anthropic says two experts judged the first version ready to send without editing.
Security and effort-level changes
Sonnet 5.5 received cybersecurity safeguards and fallback mechanisms similar to those used for Anthropic’s most powerful models. The company says the measures target a narrow set of high-risk requests, leaving ordinary software development and most life-sciences work unaffected.
Claude Code and Anthropic’s applications use a medium effort level by default, while Claude Platform uses a high setting. Sonnet 5.5 is also the first Sonnet model with classifiers designed to block extraction of its internal reasoning process.
What we know
- Anthropic has replaced Sonnet 5 with Claude Sonnet 5.5 in its mid-range Claude lineup.
- The company says the new model is more than 30% faster.
- Pricing remains $2 per million input tokens and $10 per million output tokens.
- Sonnet 5.5 was about 400 points ahead of Sonnet 5 in GDPval-AA.
- The model adds new cybersecurity safeguards and classifiers against extracting its internal reasoning process.
Trust: Single source
- The story relies on one publisher. A second independent confirmation has not been established from the cited sources.
- Cited links: 1. Publisher groups: 1. Sources: ITC.ua.
- Monitoring in the past hour: 37 of 37 RSS sources; available: 37, unavailable: 0. Matching covers published stories from the past 7 days. This does not cover the entire web.
- Additional citations require matching headlines, context, event timing and accessible article text. This automatically finds related coverage; it does not establish independence or the truth of every claim.
View sources1
COMMUNITY
Discussion
Sign in to join the discussion.
No comments yet. Start the discussion.