Overview
The upgraded Claude 3.5 Sonnet, released October 22, 2024 and often called "3.5 Sonnet (new)", was an in-place upgrade to the June 2024 model that kept the same price and speed while substantially raising capability. Its model ID is claude-3-5-sonnet-20241022.
Its headline feature was computer use: a public-beta API that lets Claude perceive a screen and control a mouse and keyboard to operate software like a person. On the OSWorld benchmark for computer use it scored 14.9% in the screenshot-only category (22.0% with more steps), well ahead of the next-best system's 7.8% at the time. On coding it raised SWE-bench Verified from 33.4% to 49.0%, and on TAU-bench it improved to 69.2% (retail) and 46.0% (airline).
This model has a 200K token context window, up to 8,192 output tokens, and supports text and vision input, at $3 / $15 per million tokens. The claude-3-5-sonnet-20241022 snapshot was retired on October 28, 2025; Anthropic recommends Sonnet 4.6 as the replacement.
| Released | 2024-10-22 |
|---|---|
| License | Proprietary |
| Weights | API only |
| Context | 200K |
| Max output | 8K |
| Architecture | Proprietary transformer (non-reasoning); upgraded Claude 3.5 Sonnet with the public-beta computer use API. |
| Knowledge cutoff | April 2024 |
| Modalities | Text, Vision, PDF |
| Status | Deprecated |
Benchmarks
Scores on a 0–100 scale (25-point gridlines); higher is better. Each benchmark links to its published source.
Pricing
| Input | $3.00 per million tokens |
|---|---|
| Output | $15.00 per million tokens |
Same price and speed as the original Claude 3.5 Sonnet.
Strengths
- Introduced computer use — controlling software via screen, mouse and keyboard
- Large coding jump: SWE-bench Verified from 33.4% to 49.0%
- Best-in-class OSWorld computer-use score at launch (14.9% / 22.0%)
- Improved agentic tool use (TAU-bench retail 69.2%, airline 46.0%)
- Same price and speed as the original 3.5 Sonnet
Best for
- Early computer-use and UI-automation agents (public beta)
- Agentic coding and software-issue resolution
- Multimodal document and screenshot understanding
- Tool-using assistants in retail and support workflows
- Form-filling and cross-application task automation
How to access
| Provider | Model ID |
|---|---|
| Anthropic ↗ | claude-3-5-sonnet-20241022 |
| Amazon Bedrock ↗ | anthropic.claude-3-5-sonnet-20241022-v2:0 |
Claude Sonnet — every version
The full lineage of the Claude Sonnet line, newest first. Every version has its own page — click any to compare specs, benchmarks and pricing.
| Version | Released | Context | License |
|---|---|---|---|
| Claude Sonnet 4.6current | 2026-02-17 | 1M | Proprietary |
| Claude Sonnet 4.5 | 2025-09-29 | 1M | Proprietary |
| Claude Sonnet 4 | 2025-05-22 | 200K | Proprietary |
| Claude 3.7 Sonnet | 2025-02-24 | 200K | Proprietary |
| Claude 3.5 Sonnet (new) | 2024-10-22 | 200K | Proprietary |
| Claude 3.5 Sonnet | 2024-06-20 | 200K | Proprietary |
| Claude 3 Sonnet | 2024-03-04 | 200K | Proprietary |
FAQ
What is the difference between the two Claude 3.5 Sonnet versions?
The October 2024 "new" version (claude-3-5-sonnet-20241022) added computer use and raised coding scores from the June 2024 original, while keeping the same name, price and speed.
What is computer use?
A public-beta API capability that lets Claude view a screen and control the mouse and keyboard to operate software, enabling UI-automation agents.
How much did coding improve?
SWE-bench Verified rose from 33.4% to 49.0%, the best of any publicly available model at the time, including reasoning models like o1-preview.
Is this model still available?
No. The claude-3-5-sonnet-20241022 snapshot was retired on October 28, 2025. Anthropic recommends Claude Sonnet 4.6.