Claude 3.5 Sonnet
One-line definition: An Anthropic model released in 2024 and retired from the Claude API in 2025.
Quick Take
- In brief: Claude 3.5 Sonnet is a retired Anthropic model name.
- Useful for: Reading older evaluations or planning an API migration.
- Main limit: Anthropic retired it from its API on October 28, 2025.
Overview
Claude 3.5 Sonnet was part of Anthropic’s 2024 model generation. It remains relevant when teams read older benchmark results or maintain software that names this model.
Core Definition
Claude 3.5 Sonnet was a general-purpose model in Anthropic’s Claude 3.5 family. Anthropic promoted it for tasks that included coding, writing, and visual input. Its former capabilities do not describe current Claude models.
Background and Evolution
Anthropic announced Claude 3.5 Sonnet in June 2024. The provider later replaced it with newer model generations. The Claude API retirement date was October 28, 2025.
How It Works
Like other language models, Claude 3.5 Sonnet processed an input context and generated an output. The application controlled the prompt, tools, and files that it supplied. The model did not independently verify that its code or explanation was correct.
Applications in Software Development and Testing
- Historical evaluation: Re-run archived tests against saved outputs or an available approved replica. Label the environment and model version.
- Migration review: Find code that refers to the old model ID. Compare a supported replacement with a representative evaluation set.
- Benchmark interpretation: Read the original task, date, and scoring method before comparing published results.
Strengths and Limitations
At release, Claude 3.5 Sonnet was a notable step in Anthropic’s model line. Historical reports can help explain earlier tool choices. The Claude API no longer serves this model. Results from it do not predict the behavior of a newer model.
Comparison with Similar Terms
| Term | Meaning |
|---|---|
| Claude 3.5 Sonnet | A retired model from Anthropic’s 2024 generation |
| Claude model family | Anthropic’s changing set of models |
| Model version | A specific model identifier with its own behavior and lifecycle |
Best Practices
- Treat the model name as a historical identifier.
- Check Anthropic’s lifecycle page before you migrate an API integration.
- Preserve the model ID, prompt, and evaluation date in historical test records.
- Run a new evaluation before you adopt a replacement model.
Common Pitfalls
Do not copy old context limits, benchmark scores, or API examples into a new integration. They can describe a retired model or an older endpoint.
FAQ
Can a team call Claude 3.5 Sonnet through the Anthropic API?
No. Anthropic lists the model as retired on October 28, 2025. Requests to a retired model fail.
Why keep this term in the encyclopedia?
The name still appears in old code, reports, and benchmark discussions. The entry explains that history and marks the API boundary.