Claude Sonnet 5.5 vs Opus 5.5: Is the Cheaper Model Now Good Enough?

What shipped

On September 28, 2026, Anthropic released Claude Sonnet 5.5, the mid-tier model in the Claude 5.5 family. It follows Opus 5.5 by six days and replaces Sonnet 5, which shipped three months earlier. TechCrunch summed up Anthropic's pitch as faster and significantly cheaper to run, and pitched at everyday work like coding and office documents.

The price per token did not move. What changed is how much work each token does. Anthropic says Sonnet 5.5 generates output more than 30% faster than Sonnet 5 and costs up to 30% less per task, because it finishes jobs in fewer tokens and fewer tool calls. It is available now in the Claude apps and on the Claude Platform, AWS, Google Cloud and Microsoft Azure, at model id claude-sonnet-5-5. Anthropic's announcement does not say which Claude app plans get it.

The benchmarks, side by side

Every number below comes from Anthropic's announcement. None has been independently reproduced yet, and The Decoder notes the cost and speed claims still need outside testing.

Two things stand out. First, on the everyday agentic work most people buy Claude for (terminals, computer use, office tasks), the gap to Opus 5.5 is one or two points, and on Terminal-Bench Sonnet is ahead. Second, the jump from Sonnet 5 on Terminal-Bench, from 10.3% to 70.6%, is so large that it is worth waiting for independent runs before planning around it. The Chartography jump, from 15.6% to 61.6%, is the same story.

Same price, cheaper jobs

The headline rate is identical to Sonnet 5: $2 per million input tokens, $10 per million output, $0.20 for cache reads and $2.50 for cache writes. Opus 5.5 costs $4 and $20, with $5 cache writes. So the savings come from efficiency, not a price cut. Anthropic says that at Low or Medium effort, Sonnet 5.5 beats Sonnet 5's best score on several benchmarks for about a tenth of the cost per task, according to VentureBeat's report.

The customer reports Anthropic published all describe the same pattern. Box says it was 2.4x faster and used 12% fewer total tokens. Zendesk says tickets were processed 20% faster. Lovable reports a third fewer tool calls and roughly half as many shell commands. Base44 says builds took 3.6 iterations on average against 7.7 for Opus 5, across 118 builds. These are hand-picked launch partners, so treat them as direction, not proof.

One practical detail: default effort is Medium in Claude Code and the Claude apps but High on the Claude Platform. If you compare Sonnet 5.5 with your current model, compare at the same effort level.

Where Opus 5.5 still wins

Anthropic itself says Opus 5.5 remains clearly stronger at difficult, open-ended work, and its own table shows it: 54.4% against 46.2% on FrontierCode 1.1, and three points ahead on Humanity's Last Exam. The Decoder also flagged an oddity at maximum effort on FrontierCode, where Sonnet 5.5 scored lower than at the setting below because it more often split work across sub-agents for code review, which sometimes caused timeouts or changes outside the task.

That matches the TechCrunch framing: Sonnet 5.5 wins on agentic coding largely because it can run several agents within the same budget, not because it reasons better than Opus on the hardest problems.

Safety changes you might notice

Sonnet 5.5 is the first Sonnet to carry the same cybersecurity safeguards as Anthropic's Opus models. In practice, higher-risk security requests fall back to Sonnet 5 rather than being answered by Sonnet 5.5. If you do legitimate security work, Anthropic points to its expanded Cyber Verification Program for tiered access. Anthropic also added classifiers meant to stop people extracting the model's reasoning to train other models, and ties preserved thinking to the account that produced it.

Who should switch

For the release that came six days earlier, see Claude Opus 5.5 and GPT-6 Sol/Luna. For the wider price landscape, see LLM pricing: complete cost comparison. The listings are at Claude Sonnet 5.5 and Claude Opus 5.5.