Claude 3.5 Sonnet
Claude 3.5 Sonnet beats Claude 3 Opus at Sonnet pricing and twice Opus's speed, and ships with the Artifacts side-panel.
Mid-tier model now outperforms the top-tier Claude 3 Opus, solving 64% of problems on Anthropic's internal agentic coding eval vs 38% for Opus. Runs at twice Opus speed, costs $3/$15 per million tokens, and has stronger vision. Remains ASL-2.
- Date
- Thursday, 20 June 2024
- Lab
- Anthropic
- Kind
- model
- Access
- closed API
- Price
- $3/$15 per M tokens in/out (2024-06-20)
Figures
| Measure | Value | Measured by |
|---|---|---|
| Internal agentic coding eval pass rate | 64% vs 38% for Claude 3 Opus | company |
| Speed vs Claude 3 Opus | 2x Anthropic claim | company |
The date is in conflict. Anthropic's own page metadata reads 2024-06-21 as captured here, while the AWS Bedrock launch post and common references give 2024-06-20, which is used here. Benchmark values (GPQA, MMLU, HumanEval) were in charts and are not listed.
Sources
- www.anthropic.com/news/claude-3-5-sonnet
- aws.amazon.com/blogs/aws/anthropics-claude-3-5-sonnet-model-now-available-in-amazon-bedroc
This record was checked against its sources on 6 October 2026. How we check