AIster · The agentic age · Products

Anthropic launches Claude 3.7 Sonnet and Claude Code

Claude Code helped make the terminal coding agent a product category

Key facts

Date

24 Feb 2025

Era

The agentic age

Interest

Products

Threads

Model wars, Agents, Labs & people

Who

Anthropic

Preview

24 Feb 2025, with Claude 3.7 Sonnet

Generally available

22 May 2025, with Claude 4

Where it runs

The terminal, later VS Code and JetBrains

One model, two ways to think

Anthropic launched Claude 3.7 Sonnet on 24 February 2025, describing it as the “first hybrid reasoning model on the market”. Anthropic put both modes into a single model: it could reply near-instantly or use extended, step-by-step thinking that the user could see. Developers could set a thinking budget of up to 128,000 tokens (the word fragments a model reads and writes). The price stayed at $3 per million input tokens and $15 per million output tokens, thinking included.

Anthropic reported 63.7% on SWE-bench Verified, a benchmark of real GitHub issues, and 70.3% with a high-compute setup.

Claude Code

The same announcement introduced Claude Code as a “limited research preview”. It is a command-line tool: a developer types `claude` in a project folder and describes a task in plain English. Anthropic said it could search and read code, edit files, write and run tests, commit and push to GitHub, and use command-line tools, keeping the developer informed along the way. In early testing, the company said, it completed tasks in a single pass “that would normally take 45+ minutes of manual work”.

The design mattered as much as the model. Claude Code worked as an agent in the developer's own environment. It had permission to act, and it worked in a loop of reading, changing and checking code.

General availability

On 22 May 2025, alongside Claude Opus 4 and Claude Sonnet 4, Anthropic announced that “Claude Code is now generally available”. It added beta extensions for VS Code and JetBrains, with proposed edits shown inline, support for background tasks through GitHub Actions, and a Claude Code SDK so developers could build their own agents on the same core. Anthropic described Opus 4 as “the world's best coding model”, scoring 72.5% on SWE-bench Verified.

Why it matters

Claude Code helped make the terminal agent a product category. Less than two months later OpenAI released Codex CLI, and through 2025 coding agents became one of the clearest commercial uses of large language models, shifting the question from how well a model can write code to how much work it can be trusted to carry out on its own.

Questions people ask

When did Claude Code come out?

Anthropic released Claude Code as a limited research preview on 24 February 2025, alongside Claude 3.7 Sonnet. It became generally available on 22 May 2025, when Anthropic launched Claude Opus 4 and Claude Sonnet 4.

What is Claude Code?

Claude Code is Anthropic's agentic coding tool. You run it in a terminal inside a project and describe a task; it can read and search code, edit files, run tests and commands, and commit changes to GitHub, asking for approval as it goes.

Sources

1. Anthropic, Claude 3.7 Sonnet and Claude Code, 2025-02-24
2. Anthropic, Introducing Claude 4, 2025-05-22
3. Anthropic, Claude Code quickstart, 2026-10-01