Anthropic releases Claude Sonnet 5.5

Background

Sonnet is one of the tiers in Anthropic's Claude naming, sitting between the Opus and Haiku models. Each tier is released in generations, and each generation gets a version number. Sonnet 5 was the previous Sonnet model, and Claude Sonnet 5.5 is the update to it. Anthropic describes Sonnet 5.5 as the second model of the Claude 5.5 family, after Opus 5.5.

What happened

On 28 September 2026 Anthropic released Claude Sonnet 5.5.^[1] Anthropic said it was its fastest Sonnet model to date.^[1] Wikipedia, citing Reuters, notes that the release came as Anthropic built toward an initial public offering (IPO), a stock-market listing in which a company sells shares to the public for the first time.^[2]

Key details

The first detail is price. Sonnet 5.5 kept the price of Sonnet 5, which is $2 per million input tokens and $10 per million output tokens.^[1] A token is the unit of text a model reads and writes. For a developer who had built a product on Sonnet 5, a model that is faster and scores higher at the same price changes the result without changing the bill per token.

The second detail is benchmark performance. On Terminal-Bench 4.0, a test of how well models complete tasks in a terminal, Sonnet 5.5 scored 70.6%. Sonnet 5 had scored 10.3% on the same test.^[1]

The third detail is safety. Sonnet 5.5 was the first Sonnet model to ship with cyber safeguards and anti-distillation classifiers.^[1] Cyber safeguards are controls meant to limit how a model can be used to attack computer systems. Anti-distillation classifiers are tools that try to detect when someone is using a model's outputs to train another model, a practice known as distillation.

Impact

The release shows the pattern of the Claude 5.5 family so far: the price stayed the same while the score and speed improved, and the safety layers were added to the middle tier of the family. For readers following the company, the timing is also part of the story. According to Wikipedia, citing Reuters, the release came as Anthropic was building toward an IPO.^[2] That context explains the company's situation at the time, but it does not change what the model does.