The AI assistant you've been using for free just got significantly smarter — and almost nobody noticed, because it happened overnight with no announcement in the app.
On June 30, 2026, Anthropic released Claude Sonnet 5 and made it the default model for every Free and Pro user on claude.ai immediately. If you opened Claude today, you were already talking to it. No settings to change, no upgrade to buy, no waitlist.
That launch puts a real question on the table for the first time in months: if you've been defaulting to ChatGPT or Gemini because Claude felt like the complicated, less accessible option, is it time to reconsider? Here's the honest comparison, with the numbers that actually matter and the one cost detail most coverage is quietly skipping over.
What Just Happened — The Short Version
💡 As of July 1, 2026: Claude Sonnet 5 is the default free model on claude.ai. It's also available on Pro ($17/month), Max, Team, and Enterprise plans, and in Claude Code. Introductory API pricing is $2 per million input tokens and $10 per million output tokens through August 31, 2026 — then it steps up to $3/$15. If you use Claude at all, you're already on Sonnet 5 without doing anything.
The strategic context matters here too. Claude Fable 5 and Mythos 5 — Anthropic's two most powerful models — are currently suspended under a US export-control order that went into effect on June 12, 2026. That makes Sonnet 5 effectively the best Claude model available to most users right now, with Opus 4.8 as the only higher-capability option still generally accessible. Anthropic timed this launch deliberately: make sure the best generally available Claude is as strong as possible while the top-tier models are offline.
How It Actually Compares to ChatGPT and Gemini
| Claude Sonnet 5 | ChatGPT (GPT-5.5) | Gemini 3.5 Flash | |
|---|---|---|---|
| Free tier available | ✅ Yes (Sonnet 5 default) | ✅ Yes (GPT-5.5 default) | ✅ Yes (3.5 Flash default) |
| Paid plan (standard) | $17/month (Pro) | $20/month (Plus) | $19.99/month (Google One AI) |
| Context window | 1 million tokens | Varies by tier | 1 million tokens |
| Agentic coding (SWE-Bench Pro) | 63.2% | 58.6% (GPT-5.5) | Not directly comparable |
| Real-time web search | Via integrations | ✅ Native | ✅ Native Google Search |
| Best known for | Agentic tasks, coding, writing | Broad general use, memory | Real-time info, Google ecosystem |
What's Actually New About Sonnet 5
The word Anthropic keeps using is "agentic" — and it's worth knowing what that actually means before deciding if it matters to you.
An agentic model doesn't just answer questions. It makes plans, uses tools like browsers and terminals, and runs through multi-step tasks autonomously without stalling halfway and asking you what to do next. The older Sonnet models were decent at coding and writing but often got stuck on tasks that required sustained, sequential decision-making — the kind of work that used to require the far more expensive Opus models.
The headline benchmark numbers from Anthropic's launch:
- SWE-Bench Pro (agentic coding): 63.2% — up from Sonnet 4.6's 58.1%, and notably higher than GPT-5.5's 58.6%
- OSWorld-Verified (computer use): 81.2% — just below Opus 4.8's 83.4%
- Terminal-Bench 2.1 (CLI/shell automation): 80.4% — which actually beats Opus 4.8's 74.6% on this specific benchmark
- Knowledge work (GDPval-AA v2): 1,618 Elo — slightly edging Opus 4.8's 1,615
✅ The bottom line on benchmarks: Sonnet 5 is the first mid-tier model to beat its Opus sibling on a major benchmark (Terminal-Bench). It doesn't beat Opus 4.8 overall — Opus still leads on the hardest coding and reasoning tasks — but the gap is now close enough that Opus is genuinely optional for most everyday use cases. For context, Sonnet 5 beats GPT-5.5 on SWE-Bench Pro coding, making it the stronger choice for developers specifically.
The Cost Detail Most Coverage Is Quietly Skipping
⚠️ If you're using the API, read this before migrating workloads. Sonnet 5 launches with introductory pricing of $2/$10 per million tokens through August 31, 2026. On September 1, it steps up to $3/$15 — a 50% per-token increase on the same sticker price as Sonnet 4.6. On top of that, Sonnet 5 uses a new tokenizer that converts the same text into roughly 1.0 to 1.35 times more tokens depending on content type. Code and structured data are hit hardest. The practical result: a workload that looks cost-neutral or cheaper today can land 20-35% above your old baseline on September's bill, even though the nominal rate card reads "unchanged." Run your actual workload through Sonnet 5 and measure the token delta before August 31 — not after.
For regular consumer users on the free or Pro plan, none of this is relevant — you're paying a flat monthly fee or nothing at all. The tokenizer caveat only matters if you're a developer calling the API by volume.
Who Should Use Which Model
Use Claude Sonnet 5 if: You do a lot of coding, writing, or multi-step task automation. You want the most capable free-tier AI for agentic work right now. You're a developer who wants near-Opus-4.8 performance at a lower API price point. You're already a Claude user — you're already on it.
Use ChatGPT (GPT-5.5) if: You rely heavily on real-time web search baked into every response. You've built habits around ChatGPT's memory and personalization features. You use plugins or GPTs that don't have a Claude equivalent.
Use Gemini 3.5 Flash if: You live in Google Workspace — Gmail, Docs, Drive, Calendar. You need the freshest real-time information from Google Search built directly into your AI responses. You're on Android and want the most integrated experience.
The honest reality in 2026 is that no single model is the best at everything. Claude Sonnet 5 is now the strongest free option for coding and agentic tasks. Gemini still wins on real-time search. ChatGPT wins on breadth of integrations and established habits. Most power users are running two of the three anyway.
Frequently Asked Questions
Is Claude Sonnet 5 free?
Yes. It became the default model for all Claude Free and Pro plan users on July 1, 2026. Free users interact with it at no cost through claude.ai, subject to free-tier usage limits.
When did Claude Sonnet 5 launch?
June 30, 2026. It became the default for free and Pro users on July 1, 2026.
Is Claude Sonnet 5 better than ChatGPT?
On agentic coding benchmarks, yes — Sonnet 5 scores 63.2% on SWE-Bench Pro versus GPT-5.5's 58.6%. For real-time web search, ChatGPT still has the edge. For general use, they're closer than they've ever been.
Is Claude Sonnet 5 better than Gemini?
It depends on the task. Sonnet 5 leads on coding and multi-step automation. Gemini 3.5 Flash leads on real-time information and Google ecosystem integration. Both are available free.
What happened to Claude Fable 5 and Mythos?
Both remain suspended under a US export-control order that took effect June 12, 2026. Sonnet 5's timing is partly a response to that — Anthropic is ensuring its strongest generally available model is as capable as possible while the top-tier models are offline.
Will Claude Sonnet 5 get more expensive?
For flat-rate subscription users, no. For API users: yes. The introductory price of $2/$10 per million tokens ends August 31, 2026, stepping up to $3/$15. A new tokenizer also means the same content generates more tokens than before, making effective cost potentially 20-35% higher from September onward.
Last updated: July 2, 2026. Benchmark data sourced from Anthropic's official Claude Sonnet 5 launch post (anthropic.com/news/claude-sonnet-5) and reporting from TechCrunch, DataCamp, and DEV Community. Pricing figures reflect Anthropic's published rates as of July 1, 2026.
0 Comments