diff --git a/api-reference/endpoint/image/edit.mdx b/api-reference/endpoint/image/edit.mdx index 3f38cdc..587861b 100644 --- a/api-reference/endpoint/image/edit.mdx +++ b/api-reference/endpoint/image/edit.mdx @@ -10,7 +10,27 @@ This is an experimental endpoint and may be subject to change. -**Pricing:** Image editing/inpainting pricing varies by model. The default model (`qwen-edit`) is **$0.04 per edit**. See the [Models endpoint](/api-reference/endpoint/models/list) for all available inpaint models and their pricing. +**Pricing:** Image editing pricing varies by model. The default model (`qwen-edit`) is **$0.04 per edit**. See the [Models endpoint](/api-reference/endpoint/models/list) for all available edit models and their pricing. + + +## Quality Tiers + +Some edit models accept an optional `quality` parameter (`low`, `medium`, `high`) that trades visual fidelity for cost. Currently supported by `gpt-image-2-edit`; other models ignore the parameter. + +```json +{ + "model": "gpt-image-2-edit", + "image": "iVBORw0KGgo...", + "prompt": "change the sky to a sunrise", + "resolution": "2K", + "quality": "medium" +} +``` + +When you omit `quality`, the model uses its default tier (`high` for `gpt-image-2-edit`). Per-tier prices (1K/2K/4K × low/medium/high) live under `model_spec.pricing.quality` in the [Models endpoint](/api-reference/endpoint/models/list) and are listed in the [Pricing overview](/overview/pricing). + + +Quality-aware pricing is the default for all API and SDK callers — no opt-in header required. Requests that omit `quality` are billed at the model's default tier (`high` for `gpt-image-2-edit`). ## Postman Collection diff --git a/api-reference/endpoint/image/generate.mdx b/api-reference/endpoint/image/generate.mdx index e8a9dc5..fac3cfd 100644 --- a/api-reference/endpoint/image/generate.mdx +++ b/api-reference/endpoint/image/generate.mdx @@ -59,6 +59,26 @@ For models that use direct pixel dimensions, send `width` and `height` instead: Check [Image Models](/models/image) or the [Models endpoint](/api-reference/endpoint/models/list) for each model's supported sizing options before mixing sizing fields across models. +## Quality Tiers + +Some models accept an optional `quality` parameter (`low`, `medium`, `high`) that trades visual fidelity for cost. Currently supported by `gpt-image-2`; other models ignore the parameter. + +```json +{ + "model": "gpt-image-2", + "prompt": "a cinematic wide shot of Venice at sunset", + "aspect_ratio": "16:9", + "resolution": "2K", + "quality": "medium" +} +``` + +When you omit `quality`, the model uses its default tier (`high` for `gpt-image-2`). Pricing depends on the combination of `resolution` and `quality` — the per-tier prices for `gpt-image-2` (1K/2K/4K × low/medium/high) are exposed under `model_spec.pricing.quality` in the [Models endpoint](/api-reference/endpoint/models/list) and listed in the [Pricing overview](/overview/pricing). + + +Quality-aware pricing is the default for all API and SDK callers — no opt-in header required. Requests that omit `quality` are billed at the model's default tier (`high` for `gpt-image-2`). + + ## Postman Collection For additional examples, please see this [Postman Collection](https://www.postman.com/veniceai/workspace/venice-ai-workspace/folder/38652128-0adc004d-2edf-4b88-a3bb-0f868c791c9c?action=share&source=copy-link&creator=38652128&ctx=documentation). diff --git a/api-reference/endpoint/image/multi-edit.mdx b/api-reference/endpoint/image/multi-edit.mdx index 098b347..075ef30 100644 --- a/api-reference/endpoint/image/multi-edit.mdx +++ b/api-reference/endpoint/image/multi-edit.mdx @@ -9,3 +9,23 @@ openapi: 'POST /image/multi-edit' **Pricing:** Multi-edit pricing varies by model. See the [Pricing overview](/overview/pricing) for current per-edit prices. + +## Quality Tiers + +Some multi-edit models accept an optional `quality` parameter (`low`, `medium`, `high`) that trades visual fidelity for cost. Currently supported by `gpt-image-2-edit`; other models ignore the parameter. + +```json +{ + "modelId": "gpt-image-2-edit", + "images": ["iVBORw0KGgo...", "iVBORw0KGgo..."], + "prompt": "blend the two scenes into one cinematic frame", + "resolution": "2K", + "quality": "medium" +} +``` + +When you omit `quality`, the model uses its default tier (`high` for `gpt-image-2-edit`). Per-tier prices (1K/2K/4K × low/medium/high) live under `model_spec.pricing.quality` in the [Models endpoint](/api-reference/endpoint/models/list) and are listed in the [Pricing overview](/overview/pricing). + + +Quality-aware pricing is the default for all API and SDK callers — no opt-in header required. Requests that omit `quality` are billed at the model's default tier (`high` for `gpt-image-2-edit`). + diff --git a/api-reference/endpoint/models/list.mdx b/api-reference/endpoint/models/list.mdx index 49648fc..19a971f 100644 --- a/api-reference/endpoint/models/list.mdx +++ b/api-reference/endpoint/models/list.mdx @@ -5,6 +5,39 @@ openapi: 'GET /models' "og:description": "Documentation covering Venice's model list API." --- +## Quality-Tier Pricing + +For image models that accept the optional `quality` parameter (currently `gpt-image-2` and `gpt-image-2-edit`), the response exposes a per-quality price matrix under `model_spec.pricing.quality`. Each top-level key is a resolution tier (`1K`, `2K`, `4K`) and each nested key is a quality level (`low`, `medium`, `high`) carrying its own `usd` and `diem` price: + +```json +"pricing": { + "resolutions": { + "1K": { "usd": 0.27, "diem": 0.27 }, + "2K": { "usd": 0.51, "diem": 0.51 }, + "4K": { "usd": 0.84, "diem": 0.84 } + }, + "quality": { + "1K": { + "low": { "usd": 0.02, "diem": 0.02 }, + "medium": { "usd": 0.07, "diem": 0.07 }, + "high": { "usd": 0.26, "diem": 0.26 } + }, + "2K": { + "low": { "usd": 0.03, "diem": 0.03 }, + "medium": { "usd": 0.13, "diem": 0.13 }, + "high": { "usd": 0.50, "diem": 0.50 } + }, + "4K": { + "low": { "usd": 0.05, "diem": 0.05 }, + "medium": { "usd": 0.21, "diem": 0.21 }, + "high": { "usd": 0.83, "diem": 0.83 } + } + } +} +``` + +`pricing.resolutions` is the legacy per-image schedule kept for backward compatibility. `pricing.quality` is the per-(resolution, quality) matrix that applies whenever the `quality` parameter is supported. Both fields are kept in the response so clients can detect quality support and surface the matrix in their own UIs. + ## Postman Collection For additional examples, please see this [Postman Collection](https://www.postman.com/veniceai/workspace/venice-ai-workspace/folder/38652128-59dfa959-7038-4cd8-b8ba-80cf09f2f026?action=share&source=copy-link&creator=38652128&ctx=documentation). diff --git a/models/image.mdx b/models/image.mdx index 25b5693..4594cfe 100644 --- a/models/image.mdx +++ b/models/image.mdx @@ -20,4 +20,8 @@ See the [Image Generate API](/api-reference/endpoint/image/generate) for text-to Image generation sizing is model-specific. Pixel-based models such as `venice-sd35` and `qwen-image` use `width` and `height`; aspect-ratio models such as `qwen-image-2` use `aspect_ratio`; models that list resolution options such as `1K`, `2K`, or `4K` use both `resolution` and `aspect_ratio`. + + + +**Quality tiers.** `gpt-image-2` (generate) and `gpt-image-2-edit` (edit/multi-edit) accept an optional `quality` parameter — `low`, `medium`, or `high` (default). Lower tiers cost less; see the [Pricing overview](/overview/pricing#quality-tier-pricing-gpt-image-2) for the full matrix or fetch `model_spec.pricing.quality` from the [Models endpoint](/api-reference/endpoint/models/list). \ No newline at end of file diff --git a/overview/pricing.mdx b/overview/pricing.mdx index ba0959f..1ec66d9 100644 --- a/overview/pricing.mdx +++ b/overview/pricing.mdx @@ -133,7 +133,7 @@ Prices per 1M tokens unless noted. All prices in USD. 1 Diem = $1/day of compute | Model | ID | Price | Privacy | |---|---|---|---| | Recraft V4 Pro | `recraft-v4-pro` | Per Image: $0.29 | Anonymized | -| GPT Image 2 | `gpt-image-2` | 1K: $0.27, 2K: $0.51, 4K: $0.84 | Anonymized | +| GPT Image 2 | `gpt-image-2` | 1K: $0.26, 2K: $0.50, 4K: $0.83 (quality `high`, default) — see [tier matrix](#quality-tier-pricing-gpt-image-2) | Anonymized | | GPT Image 1.5 | `gpt-image-1-5` | Per Image: $0.26 | Anonymized | | Nano Banana Pro | `nano-banana-pro` | 1K: $0.18, 2K: $0.23, 4K: $0.35 | Anonymized | | Nano Banana 2 | `nano-banana-2` | 1K: $0.10, 2K: $0.14, 4K: $0.19 | Anonymized | @@ -172,7 +172,7 @@ Prices per 1M tokens unless noted. All prices in USD. 1 Diem = $1/day of compute | FireRed Image Edit 1.1 | `firered-image-edit` | $0.04 | | Flux 2 Max | `flux-2-max-edit` | $0.19 | | GPT Image 1.5 | `gpt-image-1-5-edit` | $0.36 | -| GPT Image 2 | `gpt-image-2-edit` | $0.36 | +| GPT Image 2 | `gpt-image-2-edit` | 1K: $0.36, 2K: $0.53, 4K: $0.86 (quality `high`, default) — see [tier matrix](#quality-tier-pricing-gpt-image-2) | | Grok Imagine | `grok-imagine-edit` | $0.04 | | Grok Imagine High Quality | `grok-imagine-quality-edit` | $0.10 | | Nano Banana 2 | `nano-banana-2-edit` | $0.10 | @@ -184,6 +184,30 @@ Prices per 1M tokens unless noted. All prices in USD. 1 Diem = $1/day of compute | Seedream V5 Lite | `seedream-v5-lite-edit` | $0.05 | | Wan 2.7 Pro Edit | `wan-2-7-pro-edit` | $0.09 | +#### Quality-Tier Pricing (GPT Image 2) + +`gpt-image-2` and `gpt-image-2-edit` accept an optional `quality` parameter (`low`, `medium`, `high`) that combines with `resolution` to determine price. The default tier is `high`; omitting `quality` charges the `high` rate for that resolution. See the [Generate](/api-reference/endpoint/image/generate) and [Edit](/api-reference/endpoint/image/edit) docs for usage. + +**`gpt-image-2` — Generate** + +| Resolution | low | medium | high (default) | +|---|---|---|---| +| 1K | $0.02 | $0.07 | $0.26 | +| 2K | $0.03 | $0.13 | $0.50 | +| 4K | $0.05 | $0.21 | $0.83 | + +**`gpt-image-2-edit` — Edit / Multi-Edit** + +| Resolution | low | medium | high (default) | +|---|---|---|---| +| 1K | $0.03 | $0.10 | $0.36 | +| 2K | $0.04 | $0.15 | $0.53 | +| 4K | $0.06 | $0.22 | $0.86 | + + +Quality-aware pricing is the default for all API and SDK callers — no opt-in header required. Requests that omit `quality` are billed at the default tier (`high`). Programmatic clients can read the same matrix at runtime from `model_spec.pricing.quality` in the [Models endpoint](/api-reference/endpoint/models/list). + + ### Audio