Cortex 1.2
puck-cortex-1.2
- Context
- 1M
- Max output
- 256k
- Input
- $0.50 / 1M
- Output
- $1.50 / 1M
- Cached input
- $0.05 / 1M
Docs / Models
Models
One model, the same in the app and on the API. You pay per token.
How you pick#
In the app every chat runs on Cortex 1.2.
On the API, send the id in model:
| Model | API ids | Status |
|---|---|---|
| Cortex 1.2 | puck-cortex-1.2 · cortex-1.2 | Stable |
Cached tokens#
Input the model has already seen in the same chat is billed at the cached rate. On Cortex 1.2 a cached token costs a tenth of a new one, so long chats stay cheap. Your files stay on your computer: what the agent reads for the job is sent to the model to write the reply, and we keep usage for billing, not your files.
Something unclear or wrong?Tell us on Discord →