Claude Sonnet 4.5
Anthropic's mid-tier AI model released September 2025, optimized for coding and agentic tasks with a 200k token context window.
What is Claude Sonnet 4.5?
Claude Sonnet 4.5 is Anthropic's mid-tier AI model released in September 2025, built specifically for complex coding tasks and autonomous agent work.
It scored 77.2% on the SWE-bench Verified coding benchmark and can maintain focus on multi-step tasks for over 30 hours straight. The model supports a 200k token context window (with a 1M token beta version available).
Most builders use it through the Claude API, Claude.ai, or integrated into tools like Cursor. It's particularly good at understanding large codebases, building multi-file features, and running extended thinking processes where you can watch it reason through problems step-by-step.
Pricing is $3 per million input tokens and $15 per million output tokens. It sits between Claude Haiku 4.5 (faster, cheaper) and Claude Opus 4.5 (most capable, released November 2025).
Good to Know
Released September 29, 2025 as part of the Claude 4.5 family
77.2% score on SWE-bench Verified coding benchmark
200k token context window standard, 1M token beta available
Pricing: $3 input / $15 output per million tokens
Extended thinking mode shows visible reasoning process
How Vibe Coders Use Claude Sonnet 4.5
Building a multi-page feature where it edits 10+ files and keeps everything consistent
Debugging a complex issue by feeding it your entire codebase context
Watching it reason through architectural decisions in extended thinking mode
Running autonomous coding tasks that take hours without losing context
Frequently Asked Questions
Related Terms
The text instruction you give an AI model to tell it what you want it to do, like asking ChatGPT to write code or explain a concept.
The bite-sized chunks of text that AI models read and generate, like words or word fragments. They're how AI counts and processes language.
Anthropic's command-line AI assistant that lives in your terminal, reads your codebase, and writes code across multiple files.
AI agents that autonomously plan, write, test, and refactor code with minimal human prompting, handling entire workflows instead of single tasks.
The practice of crafting specific instructions to get better outputs from AI models like ChatGPT, Claude, or Gemini.
Join 0 others building with AI