Claude models
Reference for Claude model capabilities available on the platform. Use this page to pick the right model for a task and to get the correct model ID — IDs change between releases.
Current models
| Name | Model ID | Best for |
|---|---|---|
| Fable 5 | claude-fable-5 | Frontier tasks requiring the highest capability |
| Opus 4.8 | claude-opus-4-8 | Complex reasoning and long agentic work |
| Sonnet 4.6 | claude-sonnet-4-6 | Balanced workhorse — default for most tasks |
| Haiku 4.5 | claude-haiku-4-5-20251001 | Fast, cost-efficient tasks with straightforward instructions |
Practical limits
- Context window: 200K tokens
- Output: 8K tokens by default
Gotchas
- Extended thinking uses extra tokens and is noticeably slower. Only enable it for tasks that genuinely require multi-step reasoning — it is overkill for most requests.
- Haiku is fast but misses nuance on complex or multi-part instructions. Do not use it for tasks requiring careful judgment or precise instruction-following.
- Model IDs change between releases. Check this page rather than guessing — an outdated ID will fail or route to the wrong model.