Anthropic launched Claude Sonnet 5, a medium-sized model that promises performance close to that of its top model, Opus 4.8, but at a fraction of the cost. The news matters beyond the hype: it lowers the barrier to entry so that teams and companies in LATAM can run autonomous AI agents without paying premium-model prices. Here we explain what it is, how it compares, and what it means in practice.

The central fact is economic: when a capable model becomes cheap, it becomes democratized. And that changes the calculation for many SMEs and developers in the region.

What Claude Sonnet 5 is

Sonnet 5 is the new version of Anthropic's intermediate model, more powerful and "agentic" than its predecessors. According to the company itself, it can make plans, use tools like browsers and terminals, and execute tasks autonomously at a level that a few months ago required larger and more expensive models. It improves over Sonnet 4.6 in reasoning, tool use, code, and knowledge work.

How it compares to Opus 4.8

The key is in the price: at launch, Sonnet 5 starts at USD 2 per million input tokens and USD 10 per million output tokens during an introductory period, well below the high-end models. It also becomes the default model in the free and Pro plans. Anthropic also reports fewer "undesirable behaviors" like cooperation with malicious uses, something relevant for agentic contexts.

Why this brings AI agents closer to more teams

An AI agent is a system that doesn't just respond, but executes multi-step tasks: it searches for information, uses tools, and makes intermediate decisions. The problem until now was the cost: running agents for hours with a premium model was expensive. With an almost equally capable but much cheaper model, that wall falls.

For a team in LATAM, this enables cases like customer service agents, research automation, or internal assistants that operate on their own data, all with a budget that was previously prohibitive. The trend connects with what we've been seeing: when Anthropic launched Claude Opus for code and desktop tasks, it set the agentic course; Sonnet 5 makes it accessible.

You can read the full coverage in TechCrunch and the official announcement in the Anthropic blog.

What you should do with this news

If you'd been postponing an agents project because of costs, it's a good time to reevaluate. Start with a limited case (a repetitive and measurable flow), try Sonnet 5, and compare the real cost against the value it generates. The competitive advantage today doesn't belong to those who know AI, but to those who implement it first profitably.

Recommended Coderhouse courses

To take advantage of this wave of accessible agents, according to your level:

• To understand the terrain: the Introduction to Artificial Intelligence Course.

• To automate: the AI Automation Course, ideal for bringing AI to real processes.

• To build agents: the AI Agents Course, the advanced step to design autonomous systems.

Frequently asked questions

Is Sonnet 5 better than Opus 4.8?

Not in absolute terms: Opus remains the top of the line for the most complex tasks. Sonnet 5 offers performance close to it at a much lower cost, which makes it better for most practical cases.

Can I use Claude Sonnet 5 for free?

Yes, it became the default model in the free and Pro plans, with the usage limits of each plan. For intensive use via API, the price per token applies.

What exactly is an AI agent?

It's a system that executes multi-step tasks autonomously: it plans, uses tools (like a browser), and decides between actions to achieve an objective, instead of just answering a question.

Do I need to know how to program to take advantage of Sonnet 5?

For basic uses via the app, no. To build agents and integrations it does help to have technical knowledge, although there are no-code tools that reduce that barrier.