While the Claude interface looks familiar, Anthropic has completely rebuilt its engine. The new Sonnet 5.5 model delivers major performance improvements for the millions who use Claude, making it the mainstay of the 5.5 lineup.
Claude Sonnet 5.5, released yesterday, is the second model in Anthropic's 5.5 series. It is faster and more affordable than before, offering a strong alternative to the powerful Claude Opus 5.5. Built for developers and knowledge workers, Sonnet 5.5 delivers much faster speeds, better coding features, and lower costs compared to Sonnet 5.
My Workflow: Planning with Opus, Building with Sonnet
When testing these models for software engineering, I found that Sonnet 5.5 does most of the hard work. My most efficient workflow uses both models: I rely on Opus 5.5 for high-level planning and system design. After setting up the technical foundation, I pass the plans to Sonnet 5.5 for the hands-on, step-by-step building.
Sonnet's speed and cost savings make it ideal for quickly generating code, writing tests, and fixing bugs as you go.
What's New in Claude Sonnet 5.5?
The changes in Sonnet 5.5 mark a big improvement in efficiency and ability, especially for projects that take a long time.
- Over 30% Faster: Sonnet 5.5 produces results more than 30% quicker than Sonnet 5, making it the fastest Sonnet model so far.
- Up to 30% Cheaper per Task: The base token price is unchanged, but Sonnet 5.5 uses far fewer tokens to finish the same job. Since it works in fewer, stronger steps and groups tool calls well, overall costs are going down.
- Outstanding Coding Results: Sonnet 5.5 scored 70.6% on Terminal-Bench 4.0 (an agentic coding evaluation), which is a huge jump from Sonnet 5's 10.3%.
- Improved Vision Skills: The model now understands images better and became the first Sonnet model to beat Pokémon Red using only screenshots.
Pricing at a Glance
Anthropic kept the base price the same as Sonnet 5, but because Sonnet 5.5 uses tokens more efficiently, your total computing costs will probably go down.
| Token Type | Price per 1 Million Tokens |
|---|---|
| Input Tokens | $2.00 |
| Output Tokens | $10.00 |
| Cache Writes | $2.50 |
| Cache Reads | $0.20 |
The Perfect Balance
Claude Opus 5.5 is built for complex, open-ended reasoning that needs careful thought. Sonnet 5.5, on the other hand, works best in quick, repeatable tasks.
Sonnet 5.5 is especially good for:
- Agentic Coding: Quickly updating codebases, fixing bugs, and running multi-step terminal commands. Early testers at companies like Unity and Epic Games liked how it handled long tasks with fast replies.
- Knowledge Work: Creating well-formatted documents, slide decks, and spreadsheets. Sonnet 5.5 follows templates well and produces materials that need little editing.
- UI/UX Design: Testers noted its strong design skills, helping developers quickly create and improve user interfaces.
Anthropic also announced that Claude Haiku 5.5, a model made for high-volume, cost-sensitive uses, will join the 5.5 lineup soon.
How to Get Started
Claude Sonnet 5.5 is available now. Developers and business users can use it on the Claude Platform or through Amazon Bedrock, which provides AWS integration, regional data options, and strong security features.
