Models
Access 80+ AI models from OpenAI, Anthropic, Google and more through a single API, organized into budget groups.
Tchavi gives you access to 80+ AI models from OpenAI, Anthropic, Google, and more — all through a single API. Models are organized into budget groups so you can pick the right cost-to-quality tradeoff for each task.
Credit costs
Every request is billed in credits. The unit depends on the model type:
| Model type | Billing unit |
|---|---|
| Chat / text | Credits per 1K input tokens and per 1K output tokens |
| Image | Flat credits per image |
| Text-to-speech | Credits per 1K characters |
| Transcription | Credits per minute of audio |
| Video | Credits per second of output (resolution-dependent). Motion-control models reserve the maximum length upfront and refund the unused part on completion |
Browse the catalogue
The full, always-current list of models — with each model's provider, type and exact credit cost — lives in the live catalogue.
See all available models on the Models page. New models are added within 48 hours of release.
List models from the API
/v1/models/v1/models/:idOpenAI-compatible discovery. Both endpoints take your API key like any other call but are free: they consume no credits, do not count toward your rate limits and are not logged as usage. The shape matches OpenAI's /v1/models, plus a tchavi object per model with its pricing and capabilities — OpenAI clients ignore it.
{
"id": "gpt-4o-mini",
"object": "model",
"created": 1711234567,
"owned_by": "openai",
"tchavi": {
"display_name": "GPT-4o Mini",
"tier": "NANO",
"model_type": "chat",
"provider": "openai",
"context_window": 128000,
"max_output": 16384,
"pricing": { "credits_per_k_input": 1, "credits_per_k_output": 2, "credits_per_image": null },
"capabilities": {
"streaming": true,
"tools": true,
"json_mode": true,
"reference_images": false,
"documents": false,
"image_input": false
}
}
}import Tchavi from '@tchavi/sdk';
const client = new Tchavi({ apiKey: 'YOUR_API_KEY' });
const { data } = await client.models.list();
console.log(data.map((m) => m.id));
const model = await client.models.retrieve('gpt-4o-mini');
console.log(model.tchavi.pricing);Use capabilities to pick a model programmatically — for example documents: true for PDF input, or tools: true for function calling.