Anthropic Launches Claude Sonnet 5.5 With 30% Faster Output And Unchanged API Pricing

Anthropic has launched Claude Sonnet 5.5, its latest mid-tier artificial intelligence model, promising over 30% faster output, unchanged API pricing and up to 30% lower costs per task, alongside substantial gains in coding and knowledge-work benchmarks. https://www.anthropic.com/claude-sonnet-5-5

The model is the second release in the Claude 5.5 family, following Claude Opus 5.5, and is positioned as a faster and more affordable option for everyday coding, document creation, spreadsheet work and other enterprise tasks.

 

TL;DR

 

• Claude Sonnet 5.5 generates output over 30% faster than Sonnet 5, according to Anthropic.
• API pricing remains at $2 per million input tokens and $10 per million output tokens.
• Anthropic reports a 70.6% Terminal-Bench 4.0 score, versus 10.3% for Sonnet 5.
• The model also introduces stronger cybersecurity and anti-distillation safeguards.

 

Anthropic Brings Claude Sonnet 5.5 With Faster Performance

Anthropic officially unveiled Claude Sonnet 5.5 on September 28, 2026, describing it as a lower-cost complement to its recently launched Claude Opus 5.5, according to Anthropic.

According to the company, Sonnet 5.5 generates outputs more than 30% faster than Sonnet 5 and generally requires fewer tokens to complete the same work. Anthropic says this translates into task-level costs that can be up to 30% lower, even though the underlying token prices have not changed.

Sonnet 5.5 costs $2 per million input tokens and $10 per million output tokens, along with $0.20 per million tokens for cache reads. For comparison, Claude Opus 5.5 costs $4 per million input tokens and $20 per million output tokens, as detailed by Anthropic.

Reuters also reported on the unchanged pricing and Anthropic's positioning of Sonnet 5.5 as a faster and lower-cost complement to Opus 5.5.

 

Claude Sonnet 5.5 Scores 70.6% On Terminal-Bench 4.0

One of the biggest improvements Anthropic is highlighting comes in agentic coding.

The company reports that Sonnet 5.5 scored 70.6% on Terminal-Bench 4.0, compared with 10.3% for Sonnet 5. Anthropic's benchmark table lists Opus 5.5 at 66.4% on the same evaluation, according to Anthropic.

The company also reported a 55.5% score on CursorBench 4.0, up from 34.1% for Sonnet 5, while its GDPval-AA score climbed from 1,449 to 1,844. Opus 5.5 scored 1,846 on the latter evaluation.

However, Anthropic cautioned against treating those benchmark numbers as proof that Sonnet 5.5 has surpassed Opus across the board. The company says Opus 5.5 remains clearly stronger for complex, open-ended work requiring sustained judgment.

VentureBeat highlighted the model's economics in particular, noting Anthropic's claim that at some effort settings Sonnet 5.5 can outperform Sonnet 5's best benchmark results at substantially lower task costs.

TechCrunch similarly highlighted speed and reduced token usage as central selling points for the new model.

 

What Early Claude Sonnet 5.5 Testers Said

Anthropic published feedback from companies that received early access to Sonnet 5.5, including Epic Games, CodeRabbit, Slack, Zendesk, Atlassian, Unity and others.

SpaceXAI Director of ML Sualeh Asif said, “We think it will be a hit with developers looking to balance performance with cost,” according to Anthropic.

Zendesk reported that tickets in its testing were processed 20% faster, while Slack Principal Engineer Curtis Allen said Sonnet 5.5 used about 14% fewer output tokens across the company's offline Slackbot evaluations. These results come from early testers cited by Anthropic rather than independent standardized evaluations.

The Next Web also highlighted Sonnet 5.5's coding performance and the addition of safeguards previously associated with Anthropic's more capable models.

 

Claude Sonnet 5.5 Gets New Cybersecurity Safeguards

 

The performance jump has also resulted in Anthropic applying stronger safeguards to the model.

Anthropic says Sonnet 5.5 has cybersecurity capabilities comparable to Claude Opus 5, making it the first Sonnet model released with cyber safeguards similar to those used for Anthropic's most capable models. Higher-risk cybersecurity requests can fall back to Sonnet 5, while the company says routine software development should remain unaffected.

Sonnet 5.5 is also the first Sonnet release to include safety classifiers intended to prevent reasoning-extraction attacks, where large numbers of interactions can be used to extract capabilities from a model.

Its biology safeguards remain the same as those used for Sonnet 5.

 

Where Is Claude Sonnet 5.5 Available?

Claude Sonnet 5.5 is available across Anthropic's platforms as well as Amazon Web Services, Google Cloud and Microsoft Azure.

Developers using the Claude Platform can access the model through the claude-sonnet-5-5 model identifier. Anthropic also confirmed that Claude Haiku 5.5, aimed at high-volume and cost-sensitive workloads, will join the Claude 5.5 family in the coming weeks, although it did not provide a specific launch date.

The launch comes only around three months after Sonnet 5 arrived in June, underscoring the rapid pace at which Anthropic is refreshing its Claude model lineup, as also noted by TechCrunch.

The original MarkTechPost report focused on Sonnet 5.5's 70.6% Terminal-Bench 4.0 result and unchanged $2/$10 API pricing.