Anthropic just released Claude Sonnet 5.5, the second model in its Claude 5.5 family. It arrives less than a week after the company launched Claude Opus 5.5.
According to Anthropic’s announcement, Sonnet 5.5 runs more than 30 percent faster than Sonnet 5 and costs up to 30 percent less for most work. The company describes it as a lower-cost complement to Opus 5.5, and that Sonnet 5.5 is best suited to well-scoped everyday tasks, such as fixing bugs and producing documents, slides, and spreadsheets. Anthropic said Claude Haiku 5.5 will arrive in the coming weeks.
Anthropic benchmarks show big gains over Sonnet 5
Anthropic reported that Sonnet 5.5’s top score edges out Opus 5.5’s, though Opus 5.5 scored higher at every other effort level the company charted. On GDPval-AA v2.1, which measures real-world work across 44 occupations, Sonnet 5.5 scored 1,844. That is two points behind Opus 5.5 and about 400 points ahead of Sonnet 5. The company also said it is the first Sonnet model to beat Pokémon Red, working only from screenshots.
Credit: Anthropic / Mashable screenshot
Anthropic cautioned that benchmarks capture only part of a model’s abilities. The company said its own testing and outside testing both found Opus 5.5 still clearly stronger at complex, open-ended work.
Early testers cited efficiency gains. In the press release, Slack principal engineer Curtis Allen said Sonnet 5.5 beat Sonnet 5 on nearly all of Slack’s offline Slackbot evaluations, finishing “in fewer steps and with about 14% fewer output tokens.”
Pricing and availability
Pricing is the same as Sonnet 5:
-
$2 per million input tokens
-
$10 per million output tokens
-
$0.20 per million tokens for cache reads
The input and output rates are half of Opus 5.5’s. Anthropic says the lower per-task cost stems from the model requiring fewer tokens to complete the same work.
Sonnet 5.5 is available now on all of Anthropic’s platforms, including Amazon Web Services, Google Cloud, and Microsoft Azure. Developers can access it via the API as claude-sonnet-5-5, with zero data retention. Developers who run Sonnet with thinking off must switch to a new between_tools setting before migrating.
New safeguards
Anthropic said Sonnet 5.5’s cybersecurity capabilities are comparable to Opus 5’s. For that reason, it is the first Sonnet model to ship with cyber safeguards similar to those on Opus 5.5. Higher-risk cybersecurity requests will visibly fall back to Sonnet 5, while routine bug-fixing is unaffected. Biology safeguards are unchanged from Sonnet 5. Organizations can apply to the Life Sciences Verification Program for broader access. Sonnet 5.5 is also the first Sonnet model to launch with classifiers designed to block distillation attacks, in which bad actors extract a model’s capabilities at scale.
On Anthropic’s automated behavioral audit of roughly 1,850 scenarios, Sonnet 5.5 matched or improved on Sonnet 5 on most alignment measures. Opus 5.5 still performed slightly better overall. The company acknowledged that no set of evaluations catches every failure.
Source: https://mashable.com/tech/anthropic-claude-sonnet-launch-cheaper-opus-model
