The VS Code of video editing, built for agents.
TypeScript
Update @diffusionstudio/api-contract to version 0.5.0 and enhance model configurations
konstantin-paulusdaeb15f94813
diff --git a/apps/web/package.json b/apps/web/package.jsonindex b0885ca9..a8012b95 100644--- a/apps/web/package.json+++ b/apps/web/package.json@@ -14,7 +14,7 @@},"dependencies": {"@diffusionstudio/agent-chat": "*",- "@diffusionstudio/api-contract": "^0.4.0",+ "@diffusionstudio/api-contract": "^0.5.0","@diffusionstudio/assets": "*","@diffusionstudio/dapi": "*","@diffusionstudio/encoder": "*",diff --git a/apps/web/src/components/genai/config.ts b/apps/web/src/components/genai/config.tsindex 657ab392..e8bcb877 100644--- a/apps/web/src/components/genai/config.ts+++ b/apps/web/src/components/genai/config.ts@@ -12,11 +12,13 @@ import type { ModelId, Resolution } from "@diffusionstudio/api-contract";*/export const MODEL_MODES: Record<ModelId, PromptMode | "TEXT"> = {"gpt-image-2.5-sunburst": "IMAGE",+ "nano-banana-2.1": "IMAGE","nano-banana-2": "IMAGE","nano-banana-pro": "IMAGE","seedream-5.0-pro": "IMAGE","grok-imagine-image-2.0": "IMAGE","krea-2-large": "IMAGE",+ "flux-3-image": "IMAGE","flux-2-pro": "IMAGE","flux-2-klein": "IMAGE","bria-rmbg-2.0": "IMAGE",@@ -25,6 +27,7 @@ export const MODEL_MODES: Record<ModelId, PromptMode | "TEXT"> = {"kling-o3-pro": "VIDEO","wan-3.0": "VIDEO","minimax-h3": "VIDEO",+ "minimax-h3-max": "VIDEO","seedance-2.5": "VIDEO","flux-3-video": "VIDEO","grok-imagine-video-1.5": "VIDEO",@@ -34,6 +37,7 @@ export const MODEL_MODES: Record<ModelId, PromptMode | "TEXT"> = {"elevenlabs-music": "AUDIO","elevenlabs-sfx": "AUDIO","elevenlabs-v3": "VOICE",+ "elevenlabs-v4": "VOICE","gemini-3.8-flash-tts": "VOICE","universal-3.5-pro": "TEXT","qwen3.8-omni-flash": "TEXT",@@ -336,6 +340,15 @@ const video = { mode: "VIDEO" } satisfies Partial<ModelOption>;/** Most relevant first, per mode: the order the model menu lists them in. */export const MODEL_OPTIONS: ModelOption[] = [+ {+ ...image,+ id: "nano-banana-2.1",+ name: "Nano Banana 2.1",+ icon: "large-google",+ description: "Google's newest, sharper edits at Flash cost.",+ references: 14,+ resolutions: ["1K", "2K", "4K"],+ },{...image,id: "nano-banana-pro",@@ -372,6 +385,15 @@ export const MODEL_OPTIONS: ModelOption[] = [references: 14,resolutions: ["1K", "2K"],},+ {+ ...image,+ id: "flux-3-image",+ name: "FLUX.3 Image",+ icon: "large-bfl",+ description: "BFL's flagship, sharp up to 4K.",+ references: 10,+ resolutions: ["1K", "2K", "4K"],+ },{...image,id: "flux-2-pro",@@ -487,6 +509,17 @@ export const MODEL_OPTIONS: ModelOption[] = [aspectRatios: ["16:9", "9:16", "1:1", "4:3", "3:4"],resolutions: ["2K"],},+ {+ ...video,+ id: "minimax-h3-max",+ name: "MiniMax H3 Max",+ icon: "large-minimax",+ description: "Faster, cheaper H3 with native audio.",+ frames: ["start", "end"],+ durations: seconds(5, 15),+ aspectRatios: ["16:9", "9:16", "1:1", "4:3", "3:4"],+ resolutions: ["480p", "720p"],+ },{...video,id: "kling-3-pro",@@ -509,6 +542,14 @@ export const MODEL_OPTIONS: ModelOption[] = [aspectRatios: ["16:9", "9:16", "1:1", "4:3", "3:4"],resolutions: ["480p", "720p", "1080p"],},+ {+ id: "elevenlabs-v4",+ mode: "VOICE",+ name: "ElevenLabs v4",+ icon: "large-elevenlabs",+ description: "Lifelike speech, up to 10,000 characters.",+ voices: ELEVENLABS_VOICE_OPTIONS,+ },{id: "elevenlabs-v3",mode: "VOICE",diff --git a/docs/reference/models.md b/docs/reference/models.mdindex c6e1b5f3..12dcd507 100644--- a/docs/reference/models.md+++ b/docs/reference/models.md@@ -4,10 +4,10 @@ Every model [`generate`](./tools/generate.md) runs, and the fields each takes. A| Kind | Models | Makes || --- | --- | --- |-| [Image](#image) | `nano-banana-pro`, `gpt-image-2.5-sunburst`, `nano-banana-2`, `seedream-5.0-pro`, `flux-2-pro`, `grok-imagine-image-2.0`, `flux-2-klein`, `krea-2-large` | images from a prompt, or edits of reference images |-| [Video](#video) | `veo-3.1`, `veo-3.1-fast`, `kling-3-pro`, `kling-o3-pro`, `seedance-2.5`, `wan-3.0`, `minimax-h3`, `flux-3-video`, `grok-imagine-video-1.5` | a clip from a prompt, optionally from a start (and end) frame |+| [Image](#image) | `nano-banana-2.1`, `nano-banana-pro`, `gpt-image-2.5-sunburst`, `nano-banana-2`, `seedream-5.0-pro`, `flux-3-image`, `flux-2-pro`, `grok-imagine-image-2.0`, `flux-2-klein`, `krea-2-large` | images from a prompt, or edits of reference images |+| [Video](#video) | `veo-3.1`, `veo-3.1-fast`, `kling-3-pro`, `kling-o3-pro`, `seedance-2.5`, `wan-3.0`, `minimax-h3`, `minimax-h3-max`, `flux-3-video`, `grok-imagine-video-1.5` | a clip from a prompt, optionally from a start (and end) frame || [Audio](#audio) | `elevenlabs-music`, `elevenlabs-sfx` | music, sound effects |-| [Voice](#voice) | `elevenlabs-v3`, `gemini-3.8-flash-tts` | speech |+| [Voice](#voice) | `elevenlabs-v4`, `elevenlabs-v3`, `gemini-3.8-flash-tts` | speech || [Tools](#tools) | `bria-rmbg-2.0`, `seedvr-2`, `bytedance-upscaler`, `universal-3.5-pro`, `qwen3.8-omni-flash` | one file in, one file out || [Local](#local) | `sam-2.1` | runs on this machine, free: an object's mask, tracked through footage |@@ -23,10 +23,12 @@ Every model [`generate`](./tools/generate.md) runs, and the fields each takes. A| Model | | References | Aspect ratios | Resolutions || --- | --- | --- | --- | --- |+| `nano-banana-2.1` | Nano Banana 2.1 — Google's newest, sharper edits at Flash cost | 14 | 16:9, 9:16, 1:1, 4:3, 3:4 | 1K, 2K, 4K || `nano-banana-pro` | Nano Banana Pro — pro control, readable text, ultra consistent | 14 | 16:9, 9:16, 1:1, 4:3, 3:4 | 1K, 2K, 4K || `gpt-image-2.5-sunburst` | GPT Image 2.5 Sunburst — precise detail, strong text, true edits | 16 | 16:9, 9:16, 1:1, 4:3, 3:4 | 1K || `nano-banana-2` | Nano Banana 2 — fast, high quality, flexible edits | 14 | 16:9, 9:16, 1:1, 4:3, 3:4 | 1K, 2K, 4K || `seedream-5.0-pro` | Seedream 5.0 Pro — lifelike scenes, precise multi-image edits | 14 | 16:9, 9:16, 1:1, 4:3, 3:4 | 1K, 2K |+| `flux-3-image` | FLUX.3 Image — BFL's flagship, sharp up to 4K | 10 | 16:9, 9:16, 1:1, 4:3, 3:4 | 1K, 2K, 4K || `flux-2-pro` | FLUX.2 Pro — frontier quality, consistent references | 8 | 16:9, 9:16, 1:1, 4:3, 3:4 | 1K || `grok-imagine-image-2.0` | Grok Imagine 2.0 — fast, vivid, photoreal | 3 | 16:9, 9:16, 1:1, 4:3, 3:4 | 1K, 2K || `flux-2-klein` | FLUX.2 Klein — low budget, high quality, fast | 4 | 16:9, 9:16, 1:1, 4:3, 3:4 | 1K |@@ -54,6 +56,7 @@ The aspect ratios listed work on every model in its row. Some models take more (| `seedance-2.5` | Seedance 2.5 — rich motion, lip-synced audio, up to 30 s | start, end | 4–30 | 16:9, 9:16, 1:1, 4:3, 3:4 | 480p, 720p || `wan-3.0` | Wan 3.0 — expressive motion, clips up to 30 s | start | 2–30 | 16:9, 9:16, 1:1, 4:3, 3:4 | 480p, 720p, 1080p || `minimax-h3` | MiniMax H3 — native 2K with built-in audio | start, end | 5–15 | 16:9, 9:16, 1:1, 4:3, 3:4 | 2K |+| `minimax-h3-max` | MiniMax H3 Max — faster, cheaper H3 with native audio | start, end | 5–15 | 16:9, 9:16, 1:1, 4:3, 3:4 | 480p, 720p || `flux-3-video` | FLUX.3 Video — sharp detail with native audio | start, end | 5–20 | 16:9, 9:16, 1:1, 4:3, 3:4 | 720p, 1080p || `grok-imagine-video-1.5` | Grok Imagine Video 1.5 — quick clips from text or a frame | start | 1–15 | 16:9, 9:16, 1:1, 4:3, 3:4 | 480p, 720p, 1080p |@@ -75,11 +78,12 @@ Durations given as a range take any whole second in it.| Field | Type | Description || --- | --- | --- |-| `prompt` | `string`, required | the text to speak; up to 5,000 characters. ElevenLabs v3 takes audio tags in brackets (`[whispers]`, `[laughs]`); Gemini takes a style instruction before the text (`Say cheerfully: …`) |+| `prompt` | `string`, required | the text to speak; up to 10,000 characters (5,000 for ElevenLabs v3). ElevenLabs v3 takes audio tags in brackets (`[whispers]`, `[laughs]`); Gemini takes a style instruction before the text (`Say cheerfully: …`) || `voice` | `string`, required | one of the model's voices, from its voice list (linked below) || Model | | Voices || --- | --- | --- |+| `elevenlabs-v4` | ElevenLabs v4 — lifelike speech, up to 10,000 characters | [an ElevenLabs voice id](./voices/elevenlabs-v3.md) || `elevenlabs-v3` | ElevenLabs v3 — expressive speech in many voices | [an ElevenLabs voice id](./voices/elevenlabs-v3.md) || `gemini-3.8-flash-tts` | Gemini 3.8 Flash TTS — natural speech, styled in the prompt | [a Gemini voice name](./voices/gemini-3.8-flash-tts.md) |diff --git a/docs/reference/voices/elevenlabs-v3.md b/docs/reference/voices/elevenlabs-v3.mdindex 27dd476c..4fa9c6c0 100644--- a/docs/reference/voices/elevenlabs-v3.md+++ b/docs/reference/voices/elevenlabs-v3.md@@ -1,6 +1,6 @@-# ElevenLabs v3 voices+# ElevenLabs voices-The voices [`elevenlabs-v3`](../models.md#voice) speaks in: pass the id as `voice`. The prompt is the text to speak, up to 5,000 characters; audio tags in brackets steer the delivery (`[whispers]`, `[laughs]`, `[excited]`).+The voices [`elevenlabs-v3`](../models.md#voice) and [`elevenlabs-v4`](../models.md#voice) speak in: pass the id as `voice`. The prompt is the text to speak, up to 5,000 characters for v3 and 10,000 for v4; on v3, audio tags in brackets steer the delivery (`[whispers]`, `[laughs]`, `[excited]`).```json{ "model": "elevenlabs-v3", "voice": "JBFqnCBsd6RMkjVDRZzb", "prompt": "[softly] Welcome back. Let's pick up where we left off." }diff --git a/package-lock.json b/package-lock.jsonindex f622b4bb..3f7fe110 100644--- a/package-lock.json+++ b/package-lock.json@@ -92,7 +92,7 @@"license": "MPL-2.0","dependencies": {"@diffusionstudio/agent-chat": "*",- "@diffusionstudio/api-contract": "^0.4.0",+ "@diffusionstudio/api-contract": "^0.5.0","@diffusionstudio/assets": "*","@diffusionstudio/dapi": "*","@diffusionstudio/encoder": "*",@@ -774,9 +774,9 @@"link": true},"node_modules/@diffusionstudio/api-contract": {- "version": "0.4.0",- "resolved": "https://registry.npmjs.org/@diffusionstudio/api-contract/-/api-contract-0.4.0.tgz",- "integrity": "sha512-btw8Opi862XPYNxiAzx43TGj/E6SbmLnzHyuunz3LQIoaJDGXV7b31hsXhfGjeU3gnrg7bz/IFCMUdhDY1gNkA==",+ "version": "0.5.0",+ "resolved": "https://registry.npmjs.org/@diffusionstudio/api-contract/-/api-contract-0.5.0.tgz",+ "integrity": "sha512-+AZnwfitqn4V7c6RRvmcEpk1D7qvN0E4eV/Wep8sS2v+BYb5lJ3RnLMI2QYfQnLFbDEbbuxtR7nS7C/SYcZpCg==","license": "MIT","dependencies": {"@trpc/server": "^11.10.0",@@ -14183,7 +14183,7 @@"version": "0.1.0","license": "MPL-2.0","dependencies": {- "@diffusionstudio/api-contract": "^0.4.0",+ "@diffusionstudio/api-contract": "^0.5.0","@diffusionstudio/jsx": "*","zod": "^4.4.3"},diff --git a/packages/dapi/package.json b/packages/dapi/package.jsonindex c35bb514..3ec073f8 100644--- a/packages/dapi/package.json+++ b/packages/dapi/package.json@@ -18,7 +18,7 @@"zod"],"dependencies": {- "@diffusionstudio/api-contract": "^0.4.0",+ "@diffusionstudio/api-contract": "^0.5.0","@diffusionstudio/jsx": "*","zod": "^4.4.3"},