> ## Documentation Index
> Fetch the complete documentation index at: https://docs.kymaapi.com/llms.txt
> Use this file to discover all available pages before exploring further.
# Pricing Catalog
> Public pricing catalog. Single round-trip returns every active SKU across text, image, video, and audio with per-call cost detail. Cached 60s.
Public, no authentication required. Returns every **active** Kyma SKU split into two arrays:
* `text` — models billed per-token (chat / completion family). Includes `input_per_1m`, `output_per_1m`, and `cached_input_per_1m` for prompt-caching pricing.
* `media` — image / video / audio SKUs (everything non-token). Each row has `modality`, `pricing_mode`, `usd_per_call`, and a `detail` object describing the pricing shape (per-image, per-second, per-minute, per-quality, etc.).
Both arrays are sorted by `id` for stable diffs across calls. The response is `Cache-Control: public, max-age=60`.
This is the **single source of truth** for marketing surfaces, dashboard pricing tables, billing audit scripts, and the `pricing.mdx` page in these docs. If you're building a comparison page, integration tooling, or a programmatic billing audit, prefer this over scraping the docs.
```bash theme={null}
curl https://kymaapi.com/v1/pricing
```
## Response
```json theme={null}
{
"object": "pricing.catalog",
"text_count": 42,
"media_count": 19,
"text": [
{
"id": "deepseek-v3",
"name": "DeepSeek V3",
"input_per_1m": 0.27,
"output_per_1m": 1.10,
"cached_input_per_1m": 0.027
}
],
"media": [
{
"id": "veo-3",
"name": "Veo 3",
"modality": "video",
"pricing_mode": "per-second",
"usd_per_call": 0.54,
"detail": {
"provider_cost_per_sec": 0.40,
"min_duration_sec": 4,
"max_duration_sec": 8,
"allowed_durations": [4, 6, 8]
}
}
]
}
```
Always `"pricing.catalog"`.
Number of text/chat SKUs in the `text` array.
Number of image/video/audio SKUs in the `media` array.
Text models. Sorted alphabetically by `id`.
Image/video/audio SKUs. Sorted alphabetically by `id`.
### Text row fields
| Field | Type | Meaning |
| --------------------- | ------ | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| `id` | string | Kyma model ID (e.g. `deepseek-v3`) |
| `name` | string | Display name |
| `input_per_1m` | number | USD per 1M input tokens |
| `output_per_1m` | number | USD per 1M output tokens |
| `cached_input_per_1m` | number | USD per 1M cached input tokens (90% discount for models that support prompt caching). For models without caching support, this number is informational — Kyma doesn't bill at the cached rate unless your client opts in. |
### Media row fields
| Field | Type | Meaning |
| -------------- | ------ | ------------------------------------------------------------------------------------------------------------------- |
| `id` | string | Kyma model ID |
| `name` | string | Display name |
| `modality` | string | One of `image`, `video`, `audio` |
| `pricing_mode` | string | One of `per-image`, `per-second`, `per-minute`, `per-quality`, `per-mp`, `per-char`, `per-song`, `per-call` |
| `usd_per_call` | number | Representative USD cost for one typical call (e.g. 1 image at default size, 1s of video, 1 minute of audio, 1 song) |
| `detail` | object | Shape varies by `pricing_mode`. See [pricing modes](#pricing-modes) below. |
## Pricing modes
The `detail` object on `media` rows depends on the `pricing_mode`:
| `pricing_mode` | `detail` keys | Example SKUs |
| -------------- | ------------------------------------------------------------------------------------ | ------------------------------------------------------------------------------- |
| `per-image` | `provider_cost`, `markup` | `imagen-4`, `recraft-v4`, `nano-banana` |
| `per-quality` | `tiers` (object with low/medium/high → cost) | `gpt-image-2` |
| `per-mp` | `base_usd`, `extra_per_mp_usd`, `markup` | `flux-2-pro` |
| `per-second` | `provider_cost_per_sec`, `min_duration_sec`, `max_duration_sec`, `allowed_durations` | `kling-3-pro`, `veo-3`, `seedance-2-pro` |
| `per-minute` | `provider_cost_per_minute`, `markup` | `whisper-v3-turbo`, `gpt-4o-mini-transcribe-2025-12-15`, `gemini-3-flash-audio` |
| `per-char` | `provider_cost_per_kchar`, `markup` | `eleven-multilingual-v2`, `minimax-speech-hd` |
| `per-song` | `usd_per_song` | `minimax-music`, `minimax-music-pro` |
| `per-call` | `usd_per_call` | `minimax-voice-clone`, `minimax-voice-design` |
## Use cases
* **Marketing comparison pages**: fetch and render a model grid showing every SKU with current pricing, no scraping required.
* **Billing audit**: cross-check `request_logs.cost_usd` against this catalog to verify the gateway didn't drift from documented prices.
* **Dashboard model picker**: render pricing inline in the model selector without hardcoding rates client-side.
* **Programmatic SDK config**: agents that auto-select models can fetch live pricing for cost-aware routing.
## Caching
The endpoint sets `Cache-Control: public, max-age=60`. Hosted in front of Cloudflare so repeat requests within a minute hit CDN edge. Pricing changes propagate to the live response within 60 seconds of a deploy.
## Errors
This endpoint is public and read-only — no auth errors. The only failure modes are:
| Status | When |
| ------ | ----------------------------------------------------------------------------- |
| `500` | Internal error building the catalog (shouldn't happen — fail loud if it does) |
## Related
* [`GET /v1/limits/tiers`](/api-reference/limits-tiers) — companion endpoint returning the rate-limit tier matrix
* [`GET /v1/models`](/api-reference/models-list) — list models without pricing, with filtering params
* [Pricing](/pricing) — human-readable pricing page rendered from this same catalog
* [`GET /v1/credits/pricing`](/api-reference) — legacy text-only pricing endpoint (kept for backward compat; prefer this catalog)