Anthropic's Claude Sonnet 5.5 is the second model in the Claude 5.5 family. It keeps Sonnet 5's token prices, runs more than 30% faster, and Anthropic says it costs up to 30% less per task because it needs fewer tokens. Here is how it compares with Opus 5.5, where it fits best, and what it means for small businesses.
Claude Sonnet 5.5 is Anthropic's new mid-tier model, released on September 28, 2026, as the second model in the Claude 5.5 family, six days after Claude Opus 5.5. Anthropic calls it a clear upgrade over Claude Sonnet 5 that runs more than 30% faster and costs up to 30% less for most work. It keeps Sonnet 5's token prices and is available on the Claude API as claude-sonnet-5-5, plus AWS, Google Cloud and Microsoft Foundry.
Introducing Claude Sonnet 5.5, the second model in the Claude 5.5 family.
— Claude (@claudeai) September 28, 2026
It’s a clear upgrade over Sonnet 5, runs more than 30% faster, and costs up to 30% less for most work.
Post credit: video and post by Claude (@claudeai) on X, published September 28, 2026 (12.6 million views and 56,000 likes when we wrote this). View the post on X. All rights to the video belong to its creator; we embed it with X's standard embed and add our own commentary.
The post passed 12 million views in just over a week. A few reasons stand out.
Faster and cheaper at once is rare. Most upgrades trade one for the other. Sonnet 5.5 claims both, without a price increase.
The jump on agentic coding is unusually large. On Terminal-Bench 4.0, Anthropic reports Sonnet 5.5 at 70.6%, up from 10.3% for Sonnet 5. That score is also above Opus 5.5's 66.4% on the same test, so a model at half the price edges out the flagship on one widely watched coding benchmark.
It arrived right after Opus 5.5. Opus 5.5 had already reset expectations on cost. Sonnet 5.5 pushed the same story further down the lineup, and Anthropic says Claude Haiku 5.5 will follow in the coming weeks. It is also, per Anthropic, the first Sonnet model to beat Pokémon Red working only from screenshots, the kind of detail that travels well on social media.
The token prices did not change. According to Anthropic's official pricing page as of October 6, 2026:
| Price per million tokens | Claude Sonnet 5.5 | Claude Sonnet 5 |
|---|---|---|
| Input | $2 | $2 |
| Output | $10 | $10 |
| Cache reads (hits) | $0.20 | $0.20 |
| 5-minute cache writes | $2.50 | $2.50 |
| Batch API (input / output) | $1 / $5 | $1 / $5 |
The saving comes from efficiency. Anthropic says Sonnet 5.5 typically needs far fewer tokens to do the same work, and in its testing it costs up to 30% less per task. "Up to" matters: your result depends on the task, the effort setting and how much context you cache.
Two practical notes from Anthropic's developer docs. Effort levels were recalibrated, so a setting carried over from Sonnet 5 won't behave the same. And there are breaking changes for teams already on Sonnet 5, such as forced tool use no longer being supported. If someone built your agents, they will need to test before switching.
Anthropic positions Sonnet 5.5 as "a faster, lower-cost complement to Claude Opus 5.5". Opus 5.5 is built for complex work requiring careful judgment; Sonnet 5.5 is strongest at well-scoped everyday tasks. Here are Anthropic's published numbers side by side, with specs from the models overview.
| Claude Sonnet 5.5 | Claude Opus 5.5 | Claude Sonnet 5 | |
|---|---|---|---|
| API price (input / output per MTok) | $2 / $10 | $4 / $20 | $2 / $10 |
| Relative latency | Fast | Moderate | — |
| Context window | 1M tokens | 1M tokens | — |
| Terminal-Bench 4.0 (agentic coding) | 70.6% | 66.4% | 10.3% |
| FrontierCode 1.1 | 46.2% | 54.4% | 42.4% |
| CursorBench 4.0 | 55.5% | 57.8% | 34.1% |
| GDPval-AA v2.1 (knowledge work, Elo) | 1844 | 1846 | 1449 |
| OSWorld 2.1 (computer use) | 80.1% | 81.8% | 57.0% |
| Best for | Well-scoped tasks, volume, speed | Complex, judgment-heavy, long-running work | Superseded by 5.5 |
The pattern: on office-style knowledge work and computer use, Sonnet 5.5 is nearly level with Opus 5.5. On harder coding benchmarks like FrontierCode, Opus 5.5 keeps a clear lead. These are Anthropic's own results, so test on your real tasks before deciding.
Based on Anthropic's positioning and the benchmarks above, Sonnet 5.5 fits work that is clearly defined and happens often:
Reach for Opus 5.5 instead when a task needs careful judgment over many steps, such as complex research, sensitive decisions or hard multi-file code changes.
For most small businesses, Sonnet 5.5 matters more than Opus 5.5, because most useful automation is high-volume and well-defined. A missed-call follow-up, a support reply or a lead qualification email doesn't need the most powerful model. It needs a capable one that is fast and cheap enough to run on every single request.
What changes in practice:
What isn't relevant yet: if you use Claude only in a chat window, you will notice speed more than cost. And if your process isn't written down, no model will fix that. Start by defining one workflow, the approval rules and what "done" looks like. Our guide on automating customer support with AI walks through it.
Dooza is an AI-native company that builds AI products and services for small businesses, from the Dooza Workforce app to the Dooza Agents platform. Every product starts with a refundable pilot: 100% refund within 14 days.
Dooza doesn't build AI models. We build and run the agents that use them, and we choose the model per job: a fast model like Sonnet 5.5 for high-volume steps, a stronger one where judgment matters. When a better or cheaper model ships, we test it against your workflow before switching.
Dooza Workforce gives you ready-made AI employees: Maily for email, Somi for social media, Ranky for SEO and AI visibility, Stan for lead generation, Linda for legal documents, and Rachel for phone calls. Dooza Agents are custom agents built and maintained by Dooza engineers.
The best fits for a fast, efficient model are AI customer support, an AI receptionist and workflow automation. For the bigger picture, read AI agents vs agentic AI. Pricing depends on the product and is on our pricing page.
Book a free 30-minute call and a Dooza engineer will scope a pilot around one workflow. Start with a refundable pilot — 100% refund within 14 days. Book a free pilot call.
Claude Sonnet 5.5 is Anthropic's mid-tier model released on September 28, 2026, the second in the Claude 5.5 family. Anthropic says it is a clear upgrade over Sonnet 5, runs 30%+ faster and costs up to 30% less for most work.
It has the same API prices as Sonnet 5: $2 per million input tokens and $10 per million output tokens, as of October 2026. Check Anthropic's official pricing page for current rates.
Anthropic says it typically needs far fewer tokens to do the same work, so the cost per task falls by up to 30% in its testing even though the price per token is unchanged.
On Anthropic's knowledge-work and computer-use benchmarks it is within a point or two of Opus 5.5, and it scores higher on Terminal-Bench 4.0. Opus 5.5 keeps a lead on harder coding tests and complex, judgment-heavy work.
Well-scoped, frequent tasks: AI agents that run all day, everyday coding and bug fixing, customer support at volume, and documents, slides and spreadsheets.
It makes fast, capable AI agents cheaper to run on high-volume work like support replies and lead follow-up. Dooza builds and runs agents like these, each starting with a refundable pilot.
Automate your business with AI employees that work 24/7. Start with a refundable pilot: 100% refund within 14 days.
Anthropic's Claude Opus 5.5 is the first model in the Claude 5.5 family. It performs at roughly the level of Claude Fable 5.1 on most work, costs 40% less to run than Opus 5, and generates output more than 30% faster. Here is what changed, how to choose between Opus 5.5, Fable 5.1 and Sonnet 5.5, and what it means for businesses running AI agents.
A French AI account's repost of Boris Cherny's Claude Code talk reached 3.6 million views and 53,000 bookmarks. The original is Anthropic's 2025 "Mastering Claude Code in 30 minutes." Here are the actual techniques, what changed since, and how non-developers can apply them to AI employees.
Start with a refundable pilot — 100% refund within 14 days. A Dooza engineer scopes it with you on a free 30-minute call. Pricing depends on the product; see pricing.
Refundable pilot · 100% refund within 14 days · No contracts