Anthropic released Claude Sonnet 5.5 on Monday, saying its new mid-tier AI model runs more than 30% faster and costs up to 30% less per task than its predecessor. Key Points: Sonnet 5.5 score
Anthropic released Claude Sonnet 5.5 on Monday, saying its new mid-tier AI model runs more than 30% faster and costs up to 30% less per task than its predecessor.
Key Points:
- Sonnet 5.5 scored 70.6% on the Terminal-Bench 4.0 coding test, up from 10.3% for Sonnet 5.
- It is the first Sonnet model to launch with cyber safeguards once kept for Anthropic's top systems.
- One independent benchmark reportedly found it cost more per task than Sonnet 5 at maximum effort.
The model is the second in the Claude 5.5 family and arrived six days after the flagship Claude Opus 5.5, the company said in its launch announcement.
Anthropic pitched it as a faster, cheaper complement to Opus 5.5, built for well-scoped everyday tasks, bug fixes and polished documents, slides and spreadsheets. It costs $2 per million input tokens and $10 per million output tokens.
On Terminal-Bench 4.0, an agentic coding test, Sonnet 5.5 scored 70.6%, compared with 10.3% for Sonnet 5 and 66.4% for the pricier Opus 5.5. It also became the first Sonnet model to beat Pokémon Red using only screenshots, a test of long-horizon work and image understanding.
The model now runs on all Anthropic platforms, including Amazon Web Services, Google Cloud and Microsoft Azure. Anthropic said the smaller Claude Haiku 5.5, aimed at high-volume and cost-sensitive uses, will follow in the coming weeks. That would complete the three-model lineup.
Also Read:OpenAI Hits The Brakes On Frontier AI After Its Sept. 20 Sandbox Escape
Anthropic citedDaniel Vogel, chief operating officer at Epic Games, who said the model handled tens of thousands of lines of gameplay-system code in early testing. Vogel said it "cleared the same quality bar you'd expect from a higher-tier model."
Because its cyber skills are comparable to those of Opus 5, Sonnet 5.5 is the first Sonnet to launch with safeguards Anthropic had reserved for its most capable models. Routine bug fixing is unaffected, but higher-risk cyber requests visibly fall back to Sonnet 5. New classifiers also block attempts to extract its reasoning.
Not every test backs the cost claim. Artificial Analysis reportedly measured a weighted cost of $7.60 per benchmark task at maximum effort, against $5.09 for Sonnet 5, even as the new model's index score climbed to 56 from 38.
The launch follows the Sept. 22 debut of Opus 5.5, when Anthropic cut flagship pricing 20% to $4 per million input tokens and $20 per million output tokens. The company said then that Opus 5.5 would cost about 40% less than Opus 5 on typical workloads and generate output more than 30% faster. OpenAI released GPT-6 Sol and GPT-6 Luna the same day, pricing them at half the cost of their predecessors.
Read Next:BlackRock Says Computing Power Is Headed For Futures Markets
