How to Use Sonnet 4.5 in Claude

You’ve heard about Claude Sonnet 4.5. Maybe someone mentioned it on LinkedIn, or you stumbled across a benchmark that made you double-take.

And you’re probably asking yourself — do I actually need this, and how do I even use it?

Good news: it’s easier than you think. And the upgrade is genuinely worth it.

Claude Sonnet 4.5 is Anthropic’s most powerful model to date, outperforming even older flagship models across coding, reasoning, and complex tasks. It scores 77.2% on SWE-bench Verified — the gold standard for real-world coding ability — and 61.4% on OSWorld, which tests real computer interactions like navigating sites, filling spreadsheets, and completing desktop tasks.

That second number matters. Six months before Sonnet 4.5 launched, the best score on OSWorld was 42.2%. That’s a massive leap in a short time.

This article breaks down exactly how to switch to Sonnet 4.5 in Claude, what’s different about it, and whether it’s the right choice for what you’re doing. No fluff. Just what you need to know.

How to Use Sonnet 4.5 in Claude

What Is Claude Sonnet 4.5?

Think of the Claude model family like a three-tier menu.

The Claude 4.5 family includes three models: Claude Haiku 4.5 (fast and cheap), Claude Sonnet 4.5 (the balanced workhorse), and Claude Opus 4.5 (the heavy-duty flagship). Each one is built for different needs and budgets.

Sonnet 4.5 sits right in the sweet spot. It’s not the cheapest, but it punches way above its price tag. Sonnet 4.5 is priced at one-fifth the cost of Opus 4.1, while ranking in the same league as every company’s best model.

Anthropic released Claude Sonnet 4.5 on September 29, 2025. It’s available on web, iOS, Android, the Claude Developer Platform, Amazon Bedrock, and Google Cloud Vertex AI. Pricing is $3 per million input tokens and $15 per million output tokens — the same price as the previous Sonnet 4.

In other words: more performance, same price. That’s rare.

How to Select Sonnet 4.5 in Claude.ai

If you’re using Claude through the browser or app, switching to Sonnet 4.5 takes about 10 seconds.

Here’s what to do:

On Claude.ai (Web or Mobile):

1. Open Claude at claude.ai
2. Start a new conversation or open an existing one
3. Click the model selector at the top of the chat window (it usually shows the current model name)
4. From the dropdown, select Claude Sonnet 4.5
5. Start typing — you’re now using Sonnet 4.5

That’s it. No settings menus. No buried preferences. It’s front and center.

If you’re a developer using the API:

Simply use claude-sonnet-4-5 as your model string via the Claude API. Drop it in where your existing model call was, and it works as a direct replacement with no other changes needed.

Claude Sonnet 4.5 is a drop-in replacement that maintains full compatibility with existing Claude API endpoints, Amazon Bedrock, Google Cloud’s Vertex AI, and all current Claude connectors and third-party integrations. Teams can upgrade without disrupting workflows or requiring code changes in most cases.

If you’re on a Free plan, you’ll have access to Sonnet 4.5 with usage limits. Pro and Max subscribers get higher usage caps and priority access.

Performance Differences — What Actually Changed

Here’s the part that matters if you’re deciding whether to switch.

Coding and Technical Tasks

Anthropic describes Sonnet 4.5 as “the best coding model in the world” at launch, with huge gains in reasoning, mathematics, and computer use.

Internal testing shows code editing error rates fell from 9% to 0%. On AIME 2025 (advanced math), Sonnet 4.5 scores 100% with Python tools — and 87% without them.

Long Tasks and Focus

Practically speaking, Sonnet 4.5 can maintain focus for more than 30 hours on complex, multi-step tasks. That’s not a typo. Thirty hours of consistent, coherent output on a single project.

Safety and Reliability

Sonnet 4.5 achieves a 99.29% harmless response rate on violative requests (up from 98.22% for Sonnet 4), while dropping the over-refusal rate to just 0.02% (down from 0.15%). That’s a 7.5x improvement — meaning Claude is far less likely to refuse a legitimate request.

Context Window

Sonnet 4.5 supports a 200K token context window, with a 1M token option available. It can output up to 64K tokens — enough for long code generations or full technical documents.

Computer Use

On OSWorld, Sonnet 4.5 now leads at 61.4%. Just four months before, Sonnet 4 held the lead at 42.2%. The Chrome extension puts these upgraded capabilities directly into browser workflows.

Best Use Cases for Sonnet 4.5

Not every task needs Sonnet 4.5. Here’s a practical breakdown:

Use Sonnet 4.5 when you need:

• Writing and editing long documents — reports, proposals, technical specs, emails
• Complex research and analysis — synthesizing multiple sources, summarizing lengthy content
• Coding projects — building features, debugging, refactoring, writing tests
• Agent-based workflows — multi-step automation that runs over hours
• Domain-specific questions — experts in finance, law, medicine, and STEM found Sonnet 4.5 shows dramatically better domain-specific knowledge and reasoning compared to older models, including Opus 4.1.
• Sales and outreach content — writing personalized emails, LinkedIn messages, follow-up sequences at scale

Stick with Haiku 4.5 when you need:

• High-volume, simple classification tasks
• Quick one-off answers
• Cost-sensitive workflows where Sonnet-level quality isn’t required

For most applications, Claude Sonnet 4.5 offers the best blend of power, speed, and economy — making it the default choice for daily use.

Real-world results people are seeing:

• One security team reduced average vulnerability intake time by 44% while improving accuracy by 25% using Sonnet 4.5 agents.
• Developers are reporting 10× faster complex refactors and 77% higher success on real GitHub issues.
• Code editing error rates dropped from 9% to 0% on internal benchmarks.

Conclusion

Switching to Claude Sonnet 4.5 is one of the easiest upgrades you can make right now.

It’s faster than Opus 4.1, smarter than Sonnet 4, and costs the same as before. The benchmarks are strong, but more importantly — the real-world results people are seeing (zero error rates on code edits, 30+ hours of autonomous task focus) show this isn’t just marketing.

If you’re on Claude.ai, click the model selector and switch. If you’re on the API, change one line of code. That’s the whole process.

And once you’ve freed up time using tools like Sonnet 4.5, the next question becomes: how do you use that time to drive more pipeline?

That’s exactly what Salesso does. We build complete outbound systems — LinkedIn prospecting, cold email campaigns, targeting, and scaling — so you’re not just saving time, you’re converting it into booked meetings and revenue.

Book your strategy meeting with Salesso →