We currently offer two families of models:

  • Claude Instant: low-latency, high throughput.
  • Claude: superior performance on tasks that require complex reasoning.

See our pricing page for pricing details.

When making requests to our Completions API, you must specify the model to perform the completion using the model parameter. If you rely on specific output shapes from the model, we recommend that you specify the full model version (e.g. claude-2.1). If you prefer to automatically use the latest minor versions as they are released, you can specify just the major model version, e.g. claude-2.

FamilyLatest full versionLatest major version
Claude Instantclaude-instant-1.2claude-instant-1