Claude Sonnet 5 Is Now the Default Model: What Professionals Need to Know
Anthropic launched Claude Sonnet 5 on June 30, 2026 — it becomes the default for Free and Pro users on July 1. Here's what changed, what the benchmarks mean, how pricing works, and whether you need to do anything.
TL;DR. Anthropic launched Claude Sonnet 5 on June 30, 2026. It becomes the default model for Free and Pro users on July 1. Chat users need to do nothing. API users who pinned
claude-sonnet-4-6should update their model string. Sonnet 5 is substantially better than Sonnet 4.6 and roughly matches Opus 4.8 quality on most professional tasks at lower cost.
Anthropic shipped a new model today: Claude Sonnet 5. It is available immediately on all tiers — Free, Pro, Max, Team, Enterprise, Claude Code, and the API. For Free and Pro subscribers, it replaces Sonnet 4.6 as the default starting July 1, 2026.
Here is what changed, what the benchmarks actually mean, and what to do.
Where Sonnet 5 fits in the lineup
Claude's model lineup has three tiers: Haiku (fast and light), Sonnet (balanced default), and Opus (most capable). Sonnet is the model most people use day-to-day, and Sonnet 5 is the new entry at that tier.
The significant thing about this release is that Sonnet 5 narrows the gap between Sonnet and Opus substantially. On most real-world professional tasks — knowledge work, document analysis, multi-step workflows — Sonnet 5 performs at or near Opus 4.8 while costing less. That changes the practical calculus for anyone who has been paying Opus pricing for everyday work.
What actually changed
The headline improvement is agentic capability — Sonnet 5's ability to carry multi-step tasks through to completion without stalling.
Anthropic describes it as "the most agentic Sonnet yet." Concretely, it can make independent plans, use tools like browsers and terminals, and complete complex sequences of steps without losing the thread. Earlier Sonnet models would often stop mid-task when a workflow got complicated enough.
Early developer testing backs this up. One Zapier engineer said Sonnet 5 "finished end to end" on a task that "used to stall halfway." Another tester noted it "carries each one through to a tested, verified result on its own" — checking its own output without being asked.
Beyond agentic performance, Sonnet 5 also shows:
- Fewer hallucinations and less sycophancy than Sonnet 4.6 — it is less likely to confabulate details or agree with a wrong premise in your prompt
- Better prompt injection resistance — relevant if you run Claude in workflows that process external content (emails, web pages, uploaded documents)
- Improved safety accuracy — more reliable at refusing genuinely harmful requests without tripping on legitimate professional ones
The benchmarks
Three numbers worth knowing, all from Anthropic's published data:
| Benchmark | Sonnet 5 | Sonnet 4.6 | Opus 4.8 |
|---|---|---|---|
| SWE-bench Pro (coding) | 63.2% | 58.1% | 69.2% |
| Terminal-Bench 2.1 (agentic tasks) | 80.4% | 67.0% | — |
| GDPval-AA v2 (knowledge work) | 1,618 | — | 1,615 |
What these say, translated:
- Coding: Opus 4.8 still leads Sonnet 5 by about 6 points. For heavy software engineering tasks, Opus 4.8 remains the stronger choice.
- Agentic tasks: Sonnet 5 is a 13-point jump over Sonnet 4.6. This is the biggest practical improvement for professionals using Claude in multi-step workflows.
- Knowledge work: Sonnet 5 and Opus 4.8 are essentially tied. For attorneys drafting, consultants summarizing, financial professionals analyzing — the everyday tasks — Sonnet 5 is a genuine Opus alternative.
Pricing
Introductory pricing (through August 31, 2026):
- $2 per million input tokens
- $10 per million output tokens
Standard pricing (from September 1, 2026):
- $3 per million input tokens
- $15 per million output tokens
One caveat for API users: Sonnet 5 uses an updated tokenizer (the same one introduced with Opus 4.7) that generates roughly 1.0–1.35x more tokens for the same content as Sonnet 4.6. The headline token price is lower, but the tokenizer change means actual costs may be closer to what you're spending today than the per-token rate comparison suggests. Run representative inputs through the API before projecting cost changes.
For chat users on paid plans (Pro, Max, Team, Enterprise), Sonnet 5 is included in your existing subscription — no additional charge.
Who benefits most
Multi-step and automated work. If you use Claude in chained workflows — research then draft, extract then summarize, generate then check — Sonnet 5 is more reliable at completing those sequences. The agentic improvement is the clearest practical win.
Everyday professional tasks where you've been using Opus. The knowledge-work tie with Opus 4.8 means you may be able to use Sonnet pricing for tasks you've been running at Opus pricing. Worth testing before switching, but the benchmarks suggest comparable quality for most document-heavy professional work.
Single-turn tasks where you use Sonnet 4.6 already. The improvement is real but more modest — you'll notice fewer hallucinations and better follow-through on longer prompts, but the day-to-day experience for a simple "summarize this" or "draft a reply" task is not dramatically different.
What to do
Claude.ai users (Free, Pro, Max, Team, Enterprise): Nothing. Sonnet 5 becomes the default automatically on July 1. You can also switch to it manually in the model selector before then.
API users: Update your pinned model string from claude-sonnet-4-6 to claude-sonnet-5. If you're using claude-sonnet as an unversioned pointer, check Anthropic's API documentation to confirm what it now routes to.
Claude Code users: Update to claude-sonnet-5 if you had Sonnet 4.6 pinned. The agentic improvements are particularly relevant for multi-file coding tasks.
Sources
- Anthropic official announcement: Claude Sonnet 5 (June 30, 2026)
- TechCrunch: Anthropic launches Claude Sonnet 5 as a cheaper way to run agents (June 30, 2026)
- The Decoder: Anthropic's new Claude Sonnet 5 closes the gap to the pricier Opus model series (June 30, 2026)
- Anthropic System Card: Claude Sonnet 5 System Card (June 30, 2026)
Curious where AI actually fits your job?
Answer a few questions and get a free, personalized 30-day AI plan for your exact role — the tasks to automate first, and the prompts to do it.
Frequently asked questions
What is Claude Sonnet 5?+
Claude Sonnet 5 is Anthropic's newest model, launched June 30, 2026. It replaces Sonnet 4.6 as the default model for Free and Pro users starting July 1. Sonnet 5 is significantly more capable than Sonnet 4.6 — especially for multi-step agentic tasks — and approaches the quality of Opus 4.8 on many professional tasks while costing less per token. The model ID for API use is `claude-sonnet-5`.
Do I need to do anything when Claude Sonnet 5 launches?+
Probably not. If you use Claude.ai on Free or Pro, Sonnet 5 becomes your default starting July 1 automatically — no action needed. If you use the API and pinned a specific model string (`claude-sonnet-4-6`), update it to `claude-sonnet-5` to get the new model. If you used `claude-sonnet` as an unversioned pointer, check Anthropic's documentation to confirm where that pointer now routes.
How much does Claude Sonnet 5 cost on the API?+
Introductory pricing through August 31, 2026: $2 per million input tokens and $10 per million output tokens. Standard pricing from September 1: $3 per million input tokens and $15 per million output tokens. One important note: Sonnet 5 uses an updated tokenizer that generates 1.0–1.35x more tokens for the same content compared to Sonnet 4.6, so real-world costs may run slightly higher than the per-token headline suggests.
Is Claude Sonnet 5 better than Opus 4.8?+
On some tasks, yes. Sonnet 5 surpasses Opus 4.8 on real-world knowledge work benchmarks (GDPval-AA v2: 1,618 vs 1,615) and on computer use (OSWorld-Verified: 81.2%). On coding, Sonnet 5 scores 63.2% on SWE-bench Pro compared to Opus 4.8's 69.2% — so for complex software engineering, Opus 4.8 still leads. For most everyday professional tasks — document analysis, research synthesis, drafting, multi-step workflows — Sonnet 5 performs at or above Opus 4.8 level at lower cost.
What plans include Claude Sonnet 5?+
Claude Sonnet 5 is available across all plans: Free, Pro, Max, Team, and Enterprise, as well as Claude Code and the Claude Platform API. It is part of existing subscriptions — there is no Sonnet 5-specific pricing tier.
Is Sonnet 5 better for multi-step or automated work?+
Yes, meaningfully so. Anthropic describes Sonnet 5 as 'the most agentic Sonnet yet.' On Terminal-Bench 2.1 it scores 80.4% versus Sonnet 4.6's 67.0% — a 13-point jump. Early testers report it completes multi-step tasks that previously stalled, and checks its own output without being asked. If you use Claude in any workflow that chains more than a few steps — research then summarize, draft then format, extract then analyze — Sonnet 5 is more likely to carry those tasks through to completion.
Related Guides
Can Your ChatGPT History Be Subpoenaed? A Federal Case Just Changed the Picture
Federal prosecutors used a defendant's ChatGPT conversation logs as criminal evidence in a June 2026 arson trial — and a separate court ordered OpenAI to hand over 20 million chat logs in civil litigation. Here's what no AI privacy protection actually exists, and what professionals should know before typing sensitive information into any chatbot.
OpenAI's GPT-5.6 Sol Is Here — So Why Can't You Use It Yet? (June 2026)
OpenAI just launched GPT-5.6 Sol, its most powerful model yet. The catch: the Trump administration limited the rollout to around 20 government-approved partners, and regular ChatGPT users don't have access yet. Here's what GPT-5.6 Sol is, what the restriction means, and when you can expect access.
ChatGPT Just Got Better at Understanding You — and Removed Canvas (June 24, 2026)
On June 24, 2026, OpenAI updated GPT-5.5 Instant — the default ChatGPT model for all users — to better grasp what you actually mean, respond more completely to complex prompts, and adapt when you push back. Canvas is also gone from the main models. Here's what changed and what it means for your day-to-day.