Claude Sonnet (currently Sonnet 5) is Anthropic's balance of speed and capability: a 1M-token context window, image input, and a price low enough for routine use at $2/$10 per million input/output tokens. Pick it as your default for coding and writing tasks; step up to Opus only when a task genuinely needs deeper reasoning.
Other good fits
Work that benefits from a very large context window (1M tokens) without paying Opus or Fable's higher per-token price.
+ Pros
Read pros
Sonnet is Anthropic's mid-tier model: fast, a 1M-token context window, and priced for everyday use rather than occasional heavy reasoning. The current version, Sonnet 5, is described by Anthropic as "the best combination of speed and intelligence" in its lineup, and it's the model most applications should default to unless a specific task calls for something slower and more capable, or faster and cheaper.
Pick it when…
You want one default model for coding, drafting and everyday agentic tasks, without picking a different model for every task type.
Your task needs a large context window, a long document, a big codebase, without paying for Opus or Fable's higher per-token rate.
You want adaptive thinking (the model decides how much to reason) rather than manually tuning a reasoning budget.
− Cons
Read cons
On a genuinely hard, long-horizon reasoning task, a model built for depth over speed can still outperform Sonnet; don't force every hard problem through the fast default just because it's cheaper. It's a closed model with no offline or self-hosted option, so anything that must run without a network connection needs a different kind of tool entirely.
Pick something else when…
The task needs deeper, longer-horizon reasoning than Sonnet reliably gives.Claude Opus is Anthropic's next step up.
You're doing simple, high-volume classification or extraction and want the lowest per-token cost.Claude Haiku is built for that.
You need an open-weight model you can run yourself.Gemma or another open model gives you that, at a real cost in setup and hardware.
Avoid if
You need the cheapest possible model for simple, high-volume classification or extraction; [Claude Haiku](tool page) costs less per token.
Context
Sonnet 5 has a 1M-token context window and a 128K-token maximum output on the synchronous API (up to 300K on the batch API with a beta header). Its reliable knowledge cutoff is January 2026. Anthropic commits to keeping it available on its own platforms no sooner than 30 June 2027.
Requirements
Sonnet is a closed, cloud-hosted model: there's no self-hosting or offline option. Use it through the Claude API, claude.ai, or a supported cloud platform (Amazon Bedrock, Google Cloud, Microsoft Foundry), each with its own model ID and lifecycle dates.
Licence & cost
Prices checked 23 September 2026 (pricing (opens in a new tab)): $2 per million input tokens, $10 per million output tokens, through the Claude API. Batch requests run at 50% off; prompt-cache reads cost about 10% of the base input price. It's also the model behind Claude Pro's everyday usage on claude.ai. There's nothing to run or maintain; the ongoing cost is purely token usage.
First task
Documentation-based suggestion — not run here.
Give it a genuinely large input, a long report, a multi-file codebase excerpt, or a full meeting transcript, and ask for a structured summary with page or line references back to the source. This is the kind of task where Sonnet's context window matters more than its raw reasoning depth.
Done when every claim in the summary traces back to a specific part of the source, and you've caught at least one place where the model over-simplified.
Links & evidence
Documentation-based · not hands-on · Sep 23, 2026
Rests on Anthropic's models overview and pricing pages, read on 2026-09-23. No model calls were made for this review; this describes the current Sonnet generation (Sonnet 5) and will need updating when Anthropic ships the next one.
Claude models overview · Sep 23, 2026 — Sonnet 5 specs: 1M context, 128K max output, January 2026 knowledge cutoff, retirement not before June 2027.
Claude API pricing · Sep 23, 2026 — Sonnet 5 at $2/$10 per million input/output tokens; batch and cache-read discounts.
Choosing a model · Sep 23, 2026 — Guidance on when Sonnet is the right default versus Opus or Fable.
Further review notes & alternatives
Alternatives and trade-offs
Claude Opus wins when the task needs more reasoning depth than speed.
Claude Haiku wins when the task is simple and high-volume, and cost per call matters most.
GPT-5 wins as the closest cross-vendor comparison; worth benchmarking both on your actual task before committing.
Handle everyday coding, writing, and agentic tasks with a 1M-token context.
Choose it when image input, caching, and batch processing fit production work.
Its closed cloud deployment supports routing through selected cloud platforms.