Claude Sonnet 4.6 Explained: What Actually Changed
Anthropic released Sonnet 4.6 on February 17, 2026, and quietly made it the default model for every Free and Pro Claude user the same day. If you haven't picked a model manually since then, you've already been using it. Here's what changed under the hood, and whether it's worth paying attention to.
Sonnet 4.6 is now the default model behind claude.ai for anyone on the Free or Pro plan.
A Full Refresh, Not a Minor Patch
Sonnet 4.6 is Anthropic's latest mid-tier model, and the company describes it as a full upgrade across coding, computer use, long-context reasoning, agent planning, knowledge work, and design — not an incremental tweak to Sonnet 4.5. It's available on claude.ai, Claude Cowork, Claude Code, the API, and major cloud platforms including Amazon Bedrock, Google Cloud's Vertex AI, and Microsoft Foundry.
The most notable technical change is a 1 million token context window, now in beta, up from Sonnet 4.5's 200,000. Anthropic frames that as enough to hold entire codebases, lengthy contracts, or dozens of research papers in a single request — and says the model reasons effectively across all of it, not just accepts the larger input without meaningfully using it.
The Improvements That Actually Matter
The 1M token context window is large enough to hold entire codebases in a single request, per Anthropic.
Improved computer use lets Sonnet 4.6 navigate browsers and automate multi-step desktop workflows more reliably.
1M Token Context (Beta)
Holds entire codebases, lengthy contracts, or dozens of research papers in one request — currently available on the API only, not yet in the consumer chat interface.
Stronger Coding
Anthropic says early-access developers preferred Sonnet 4.6 to its predecessor "by a wide margin," and often even to Opus 4.5, the company's smartest model as of November 2025.
Better Computer Use
Improved ability to navigate browsers, fill multi-step web forms, and automate desktop workflows — the same capability that underpins tools like Claude Cowork.
Context Compaction (Beta)
Automatically summarizes older parts of a long conversation as it approaches system limits, helping sustain coherence over extended agentic sessions.
Sonnet 4.6 vs Opus 4.6 vs Sonnet 4.5
| Model | Context Window | Best For | Price (per 1M tokens) |
|---|---|---|---|
| Sonnet 4.6 | 1M (beta) | Daily coding, agents, most knowledge work | $3 in / $15 out |
| Opus 4.6 | 200K | The hardest reasoning and planning tasks | Higher, premium tier |
| Sonnet 4.5 | 200K (1M available separately) | Superseded by 4.6 as the default | $3 in / $15 out |
Anthropic's own benchmarking claims Sonnet 4.6 beats Opus 4.6 on some agentic financial analysis and office-task categories, which is a notable claim given Opus is the more expensive, higher-tier model. That doesn't mean Sonnet 4.6 replaces Opus everywhere — for the very hardest, most open-ended reasoning problems, Opus-class models still tend to lead.
What This Release Does and Doesn't Change for You
- No action needed — Free and Pro users were switched over automatically
- Meaningfully better coding consistency and instruction-following, per Anthropic's early-access testers
- Same pricing as the previous Sonnet generation, so no cost increase for API users
- Available everywhere Claude already is: web, mobile, Cowork, Code, and major cloud platforms
- The 1M token context window is API-only and still in beta, not yet in the standard chat interface
- Opus-class models still lead on the single hardest reasoning benchmarks
- Benchmark claims come from Anthropic's own testing, not fully independent verification
- An updated Haiku model was expected to follow, so the lineup may shift again soon
Already Upgraded, Genuinely Worth Noticing
The rare model update that closes the gap to the flagship tier, not just improves on itself.
Most model updates are incremental. Sonnet 4.6 is notable because Anthropic is explicitly positioning it as covering ground that used to require the more expensive Opus tier — for a mid-tier model, that's a meaningful claim, even accounting for the fact it comes from Anthropic's own benchmarking. If you're a Free or Pro user, there's nothing to do; you're already on it. If you're deciding between plans or models, the practical takeaway is that Sonnet's ceiling moved up without its price changing.
Frequently Asked Questions
Do I need to do anything to start using Claude Sonnet 4.6?
No. Anthropic made Sonnet 4.6 the default model for Free and Pro users on claude.ai and Claude Cowork automatically. If you've used Claude any time since February 17, 2026, you've likely already been using it without needing to pick it manually.
Is Sonnet 4.6 better than Opus?
For many everyday and coding tasks, Anthropic says early testers preferred Sonnet 4.6 over the smartest model available before it, Opus 4.5, and that it can be faster than the newer Opus 4.6 in many cases. Opus-class models still lead on the very hardest reasoning tasks, but the gap has narrowed enough that Sonnet 4.6 covers ground that used to require paying for Opus.
How much does Claude Sonnet 4.6 cost to use?
Pricing is unchanged from the previous Sonnet model: $3 per million input tokens and $15 per million output tokens through the API, with up to 90% savings available through prompt caching and 50% through batch processing. It's included at no extra cost within Claude's Free and Pro consumer plans, which start at $20 a month for Pro.
More AI Model Guides, Written Clearly
Browse our full library of AI model explainers, updated as new releases land.
More from the creator
Explore two other projects from the same creator.