Claude Sonnet 5.5: Features, Pricing & vs Opus (2026)

Last Updated: November 1, 2026

Summary: Claude Sonnet 5.5, released September 28, 2026, is Anthropic’s newest mid-tier model and the current version of Claude Sonnet. It costs $2 per million input tokens and $10 per million output tokens, the same as Sonnet 5 and half the price of Opus 5.5. Anthropic reports it beats Opus 5.5 on Terminal-Bench 4.0 and essentially ties it on knowledge work, while running more than 30% faster than Sonnet 5.

If you've opened Claude recently, there's a good chance Sonnet is the model answering you. Claude Sonnet is Anthropic's mid-tier model line, sitting between the fast, low-cost Haiku and the more capable Opus. The current version, Claude Sonnet 5.5, launched September 28, 2026, and it beats Claude Opus 5.5 on Terminal-Bench 4.0 and comes within a few points of it on several other benchmarks while costing half as much. This guide covers what Claude Sonnet is, what changed in Sonnet 5.5, how it compares to Opus 5.5 and Haiku 5.5, what it costs, and how the naming evolved from 3.5 and 3.7 to 5 and now 5.5.

💡 Not sure which AI tool is actually right for your business?

Get the free guide, Which AI Tool Should You Actually Use?, a straightforward breakdown of the leading AI tools to help you pick the right one for your needs. Subscribe to AI Business Weekly for the guide, plus daily coverage of AI trends, acquisitions, and product launches.

Get the Free Guide →

Claude Sonnet is the middle tier in Anthropic’s three-model lineup, positioned between Claude Haiku (fastest, cheapest) and Claude Opus (most capable, most expensive). The current version, Claude Sonnet 5.5, is built for agentic work: making plans, using tools like browsers and terminals, and completing multi-step tasks on its own.

Sonnet 5.5 replaced Sonnet 5, which launched June 30, 2026. Anyone can chat with Sonnet 5.5 on Claude.ai, on the web and in the iOS and Android apps. Developers can use it through the Claude API as claude-sonnet-5-5, and on AWS, Google Cloud, and Microsoft Azure.

Claude Sonnet 5.5 is a big jump over Sonnet 5, and on several tests it now matches or beats Claude Opus 5.5. Anthropic's Sonnet 5.5 launch post and models documentation list these changes:

Sonnet 5.5 benchmarks (reported by Anthropic)

Independent testing from Artificial Analysis also put Sonnet 5.5 ahead of Opus 5.5 on Terminal-Bench 4.0 (63.6% vs 59.6%). One catch: at max effort, Sonnet 5.5 can use more tokens than Opus 5.5 on hard tasks, so the cheapest setting isn’t always max.

Claude Sonnet has five versions people still search for: Claude 3.5 Sonnet, Claude 3.7 Sonnet, Claude Sonnet 4.6, Claude Sonnet 5, and the current Claude Sonnet 5.5, released September 28, 2026. Claude 3.5 Sonnet and Claude 3.7 Sonnet still get meaningful search volume after being superseded, because plenty of documentation, integrations, and older comparisons reference them by name.

Anthropic moved from names like "Claude 3.7 Sonnet" to "Claude Sonnet 4.6," putting the tier name before the version number, and kept that format for Claude Sonnet 5 and Claude Sonnet 5.5. That switch is part of why searches for the older names haven't died down: people still search using the naming pattern they learned first. If you're troubleshooting an older integration, a tool or tutorial that says "Claude 3.5 Sonnet" is describing an earlier-generation model, not a typo or a different product line.

The honest answer: the gap is almost gone. Anthropic reports Sonnet 5.5 scoring 70.6% on Terminal-Bench 4.0 versus 66.4% for Opus 5.5, and on GDPval-AA, a test of real-world work tasks, the two essentially tie (1844 vs 1846). Opus 5.5 still leads slightly on computer use and coding inside Cursor, but Sonnet 5.5 costs half as much.

The gap between Sonnet and Opus closed fast. When Sonnet 5 launched in June 2026, TechCrunch's testing found it trailing Opus on agentic coding, 63.2% to 69.2%. Anthropic released Claude Opus 5.5 on September 22, 2026, according to the Anthropic newsroom. Six days later, Claude Sonnet 5.5 arrived and brought the mid-tier model level with Opus on most of Anthropic's own tests.

Claude 5.5 family at a glance (API list prices per million tokens, checked November 1, 2026)

Source: Anthropic's Claude pricing page and models overview.

“Agentic” gets used loosely across the AI industry, so it’s worth being specific about what it means for Sonnet 5.5. Instead of answering one prompt and waiting for the next instruction, Sonnet 5.5 takes a goal, breaks it into steps, uses tools like a browser or a terminal to carry out those steps, and checks its own work before handing back a result. Early users of the Sonnet line, cited in TechCrunch’s coverage of the Sonnet 5 launch, noted the model “completes complex multi-step tasks end-to-end” and “checks its own output without explicitly being asked.” That’s a real shift from a model that simply answers whatever the last message asked.

In practice, Sonnet 5.5's agentic design shows up most clearly in coding and research. Instead of writing one function and stopping, Sonnet 5.5 can plan a change across several files, make it, run tests, and fix its own mistakes in a single task. Coding is the work where Sonnet 5.5 has caught up to Opus the most, while Opus still has a small edge on tasks that need deep, careful judgment across a very long context. For how Claude models perform inside real coding tools, see our Claude Code vs Cursor comparison.

Sonnet's practical sweet spot is high-volume, moderate-complexity work where cost compounds fast. Developers use it for day-to-day coding: writing functions, reviewing pull requests, and running multi-step debugging sessions where an agentic loop saves time over manual back-and-forth. Content and research teams use it for drafting, summarizing, and multi-source synthesis where the task is substantial but not at the level of a full investigative report. Businesses running Claude through the API for customer-facing tools, such as support triage, internal documentation search, and structured data extraction, tend to default to Sonnet because per-token cost matters at volume in a way it doesn't for a single one-off task.

The common thread across all of these is repetition and scale. A task run once a month can absorb Opus pricing without much thought; a task run thousands of times a day cannot, which is exactly the gap Sonnet is built to close.

Claude Sonnet 5.5 costs $2 per million input tokens and $10 per million output tokens on the Claude API, according to Anthropic's pricing page. That is the same list price as Sonnet 5 and half the price of Claude Opus 5.5 ($4/$20). Sonnet 5.5 cache reads cost $0.10 per million tokens, half of Sonnet 5's $0.20, and the Batch API cuts input and output prices by 50%, to $1 and $5. Anthropic says the bigger savings come from efficiency rather than a price cut: Sonnet 5.5 costs up to 30% less per task than Sonnet 5 because it uses fewer tokens.

For individual users, Claude Sonnet 5.5 is available to anyone chatting on Claude.ai on the web and in the iOS and Android apps, so per-token API pricing mainly matters for developers building on Claude directly. Our Claude API pricing guide breaks down every model and discount in more detail.

Take a workload of 10,000 requests a month, each using 20,000 input tokens and 2,000 output tokens. That is 200 million input tokens and 20 million output tokens a month.

Put simply, picking Opus 5.5 over Sonnet 5.5 for this workload costs an extra $600 a month, or $7,200 a year, for a benchmark edge Anthropic measures in single digits. One caveat on token counts: Anthropic's pricing page notes that Claude 4.7 and later models use a newer tokenizer that produces about 30% more tokens for the same text than Sonnet 4.6 and earlier, so budgets carried over from older models can run low.

Sonnet 5.5 is now the better fit for most work: coding, drafting and editing, research summaries, and any agentic workflow where you run many tasks and cost adds up. Opus 5.5 still earns its higher price on the hardest jobs, like a codebase-wide migration or a deep audit where every step needs careful judgment. With the benchmark gap this small, most teams should start on Sonnet 5.5 and switch to Opus only when a task keeps failing.

Anthropic's own models overview recommends starting with Claude Opus 5.5 for most workloads and using Claude Fable 5.1 for demanding reasoning and long-horizon agentic work. For cost-sensitive teams, the benchmarks point the other way: Claude Sonnet 5.5 trails Opus 5.5 by less than 3 points on OSWorld 2.1 and CursorBench 4.0 at half the price, which makes it a sensible default for well-scoped, high-volume work. Claude Haiku 5.5, at $0.10 input and $0.50 output per million tokens, fits simpler, very high-volume tasks. For a broader look at how Claude compares to other AI assistants, our ChatGPT vs Claude and Claude vs Perplexity comparisons cover that ground.

Claude Sonnet is Anthropic's agentic, mid-tier model, and with Sonnet 5.5 it has nearly caught up to Opus. Released September 28, 2026 at $2/$10 per million tokens, Sonnet 5.5 matches or beats Opus 5.5 on several of Anthropic's benchmarks, runs more than 30% faster than Sonnet 5, and costs half as much as Opus. For most everyday work, Sonnet 5.5 is the practical choice. Opus 5.5 is worth the premium mainly for the longest, hardest jobs.

What is Claude Sonnet used for?

Claude Sonnet 5.5 is built for well-scoped coding and knowledge work: writing and reviewing code, planning and completing multi-step tasks, using tools like browsers and terminals, and drafting or research. Anthropic describes it as a hybrid reasoning model with fast, cost-efficient intelligence and positions it as a lower-cost complement to Claude Opus 5.5. The hardest, longest jobs still favor Opus 5.5 or Fable 5.1.

Is Claude Sonnet better than Claude Opus?

Not better overall, but close on almost every test and half the price. Anthropic’s data shows Sonnet 5.5 matching Opus 5.5 on knowledge work (1844 vs 1846 on GDPval-AA) and beating it on Terminal-Bench 4.0 (70.6% vs 66.4%). Opus 5.5 still leads slightly on computer use and on the longest, most complex jobs, like a full codebase migration. If cost matters and the task isn’t extreme, Sonnet 5.5 is the better starting point.

How much does Claude Sonnet cost?

Claude Sonnet 5.5 costs $2 per million input tokens and $10 per million output tokens, the same as Sonnet 5 and half the price of Opus 5.5 ($4/$20). The Batch API halves that to $1 and $5, and cache reads cost $0.10 per million tokens. Most individual users access Sonnet through Claude.ai instead of paying per token; see our Claude API pricing guide for the full breakdown.

What happened to Claude 3.5 and 3.7 Sonnet?

Claude 3.5 Sonnet and Claude 3.7 Sonnet are older generations that Anthropic has replaced, first with Claude Sonnet 4.6, then Sonnet 5 in June 2026, and Sonnet 5.5 on September 28, 2026. Neither appears on Anthropic's current models page, though some third-party tools still reference them. If you're building something new, Claude Sonnet 5.5 is the version to target.

Is Claude Sonnet 5.5 the newest Claude model?

Claude Sonnet 5.5, released September 28, 2026, is the current Sonnet model and replaced Sonnet 5. Anthropic's newest release is Claude Haiku 5.5 (October 7, 2026), and its most capable model is Claude Fable 5.1 at $10/$50 per million tokens. Claude Sonnet 5.5 sits in the middle at $2/$10, with a 1 million token context window.

By Sameer Khan

This article was AI-assisted, then reviewed by Sameer Khan before publishing.

Sameer Khan is the founder of AI Business Weekly. He has a background in research and advisory, working with HR leaders and executives across Canadian public-sector and enterprise organizations on research and AI adoption. He holds an MBA from the Ted Rogers School of Management and has spent nearly a decade in B2B sales across SaaS, research and advisory, and AI.