AI Model: Claude 3.7 Sonnet (anthropic)
1 → partial
Our claim
entire record- Entity Type
- ai-model
- Title
- Claude 3.7 Sonnet
- Description
- Claude 3.7 Sonnet was released February 24, 2025 as the first hybrid reasoning model. Users could toggle extended thinking mode on or off, allowing the model to self-reflect before answering. Priced at \$3/\$15 per million tokens with a 200K token context window.
- Tags
- claude,reasoning,extended-thinking
- Last Updated
- February 2025
- Related Entries
- [object Object],[object Object]
- Metadata
- [object Object]
Source evidence
1 src · 1 checkNoteThe source confirms title (Claude 3.7 Sonnet), developer (Anthropic), release date (Feb 24, 2025), input price ($3), and output price ($15). However, three fields are problematic: (1) Context window claimed as 200,000 tokens, but source only mentions '128K tokens' as the output limit/thinking budget ceiling—the input context window is not explicitly stated in the source. (2) Safety level claimed as 'ASL-2' is not mentioned anywhere in the source; the source discusses safety improvements and a system card but does not use the ASL-2 designation. (3) The pricing values match exactly when converted to per-token rates ($3 per million = $0.000003 per token; $15 per million = $0.000015 per token), confirming those fields. The record is partially confirmed because most key fields are accurate, but the context window and safety level designations cannot be verified from this source.