Anthropic has a new model on the block, and it is not the one everyone expected to lead the charge. Sonnet 5.5 arrived today, and the company says it runs 30 percent faster than its predecessor while costing significantly less. The twist is that it still sits below the company’s flagship Opus model.
The announcement comes amid a crowded field. OpenAI released a number of new models last week, including upgraded versions of Sol and Luna, its mid-tier and budget-friendly offerings. Meta also dropped word of a new model tied to its smart glasses. The pace of releases shows no sign of slowing.
What Sonnet 5.5 Does
Sonnet 5.5 is pitched as an everyday workhorse. Anthropic says it handles coding, office documents and other routine tasks. The lab describes it as an ideal assistant for those jobs.
The model’s predecessor, Sonnet 5, was announced about three months ago. Its main claim was efficient agentic deployment — running agents at a lower cost than competitors. Sonnet 5.5 pushes past that into speed. Anthropic says its rate of token burn is significantly slower too.
“30 percent faster than its predecessor”
The comparison to Opus is where things get interesting. In Anthropic’s own hierarchy, Sonnet is less powerful than Opus. But the company says Sonnet 5.5 can be more useful in certain circumstances because of its agility. Benchmarks show Sonnet 5.5 performing better than Opus 5.5 on agentic coding, likely because it can spawn multiple agents without exceeding cost limits.
Cyber Safeguards and the Haiku Plan
Cyber capabilities are also part of the pitch. Anthropic claims Sonnet 5.5 has “comparable” cyber capabilities to Opus 5. That means 5.5 will be the first Sonnet model subject to the same cyber safeguards that apply to Fable and Opus.
The company also plans to release a new version of Haiku, its smallest model, in the coming weeks. No firm date was given.
What This Means for Users
The pattern holds across the industry. Labs keep pushing models downmarket, each one cheaper and faster than the last. Anthropic’s move follows that script, but with a notable wrinkle: the model below the top dog is now the one doing the heavy lifting.
The company’s own framing is that agility trumps raw might. Sonnet 5.5 trades pure power for efficiency, and Anthropic is betting users will notice the difference in daily tasks.
The coming weeks will test that bet. Haiku’s upgrade arrives soon, and OpenAI’s Sol and Luna upgrades are already live. The model wars are far from over.
Key Facts Box
- Sonnet 5.5: 30 percent faster than predecessor
- Token burn: significantly slower than predecessor
- Cyber capabilities: comparable to Opus 5
- Haiku update: coming in the coming weeks, no firm date
- OpenAI: released Sol and Luna upgrades last week
- Meta: announced a new model for smart glasses
Anthropic’s move is a reminder that speed and cost are the new measure of success. Whether Sonnet 5.5 can hold its own against the pack remains to be seen.
Source material: “Anthropic releases Sonnet 5.5, which it calls a significantly cheaper, faster work partner,” TechCrunch.
Get the Notebook.
The day's best stories and every fresh verdict, in plain English, in your inbox by seven. One email a day, no more.