Models and capabilities

MoonNexAI provides OpenAI-compatible, multimodal, video, music, digital human and asset APIs. Public documentation describes the model IDs, task APIs and asset APIs available to users.

Capacity	Annotations	Main entrance
OpenAI-compatible	Dialogue, Responses, Messages, Embeddings and Audio Basic Capability	/v1/*
Picture	Photo generation, editing, variant and asynchronous pictures task	/v1/images/*
Video	Video generation, query, cancellation, content reading, extension and remix	/v1/videos, /v1/videos/{task_id}
Digital humans	Avatar assets, voice assets, speech and digital human video synthesis	/v1/assets/uploads, /v1/tasks
Music	Music generation, lyrics generation, sound, editing and reprocessing	/v1/tasks
Asset resources	Upload, register, group and reuse media assets	/v1/assets/*, /v1/asset-groups/*

Model task

The model task is usually submitted via POST /v1/tasks or the corresponding capacity portal. Once submitted, task_id is saved and then the state is read with GET /v1/tasks/{task_id} or a capability-specific query API.

This new model portal

• Photo asynchronous model: seedream-5.0, seedream-4.5, wan2.7-image-pro; API Key, which has an asynchronous capability, can also use gemini-3-pro-image-preview. Access is /v1/images/generations/async, query is /v1/images/tasks/{task_id}.
• Video models: happyhorse-1.0, wan2.7, entry /v1/videos, query /v1/tasks/{task_id} or /v1/videos/{task_id}.
• Seedance SG Series: seedance-2.0-sg-mini, seedance-2.0-sg, seedance-2.0-sg-fast, seedance-2.5-sg, supporting generic and official compatible parameters, single graphs and hybrid references, settled by output Token; 2.0 Standard version supporting 480P / 720P / 1080P / 4K, Fast/Mini supporting 480P / 720P, 2.5 supporting 480P / 720P / 1080P.
• Seedance NT: seedance-2.0-nt, seedance-2.0-nt-fast, seedance-2.0-nt-mini and seedance-2.5-nt, via /v1/videos or /api/v3/contents/generations/tasks. Only 720P is supported. 2.0 Standard/Fast support 4–15 seconds, Mini only 4 seconds, and 2.5 4–30 seconds. Settlement uses actual usage.total_tokens.
• Seedance-2 CL Lightweight Model: seedance-2.0-cl-mini, entry /v1/videos, supports 480P / 720P, 4 - 15 sec, text-to-video, image-to-video, Video Reference and this Model Line Asset:// media assets.
• Seedance 2.5 CL: seedance-2.5-cl, via /v1/videos, supports 4–30 seconds, 480P / 720P / 1080P, mixed image/video/audio references and authenticated Asset:// assets prepared for this model. Failed asset resolution returns an error without falling back to a public URL.
• The Seedance-2 CL-B series: seedance-2.0-cl-b-fast, seedance-2.0-cl-b-mini, entry /v1/videos, which is compatible with the official shape of /api/v3/contents/generations/tasks; parameters, media assets type and capacity boundary are consistent with the Volcengine-compatible Seedance 2.0 Fast/ Mini Pact, respectively.
• Seedance-2 SZ Series: seedance-2.0-sz, seedance-2.0-sz-fast, seedance-2.0-sz-mini, seedance-2.5-sz, entry /v1/videos; support for photo, video and audio reference, based on the Seedance official Token scale billing, according to model list API.
• Seedance-2GL series: seedance-2.0-gl, seedance-2.0-gl-fast, seedance-2.0-gl-mini, entry /v1/videos. Three models support up to 9 pictures, 3 videos, 3 audio, reference media assets totals up to 15, supporting real people; standard versions support 1080P, Fast and Mini support 480P/720P.
• Seedance-2 Original Series: doubao-seedance-2-0-260128, doubao-seedance-2-0-fast-260128, doubao-seedance-2-0-mini-260615, entry /v1/videos, supporting 4 - 15 seconds and graphic, video, audio reference; standard version supporting 1080P, Fast and Mini supporting 480P/720P.
• Moon-2 WC 720P: moon-2.0-wc-720p, moon-2.0-wc-b-720p, moon-2.0-wc-b-720p-fast, moon-2.0-wc-c-720p, moon-2.0-wc-d-720p, moon-2.0-wc-e-720p and moon-2.0-wc-f-720p, via /v1/videos. All seven support real people. Base, B and B Fast accept up to 9 images + 3 videos + 3 audio files, with at most 12 assets total; B/B Fast do not support text-only input. F supports text-only input and 4–15 seconds, with the same per-type and total limits, and requires an image when audio is present. C supports text-only input or up to 9 images, without video/audio. D/E support text-only input or up to 4 images + 3 videos + 1 audio file; audio-only input is prohibited.
• Moon-2 GL 720P: moon-2.0-gl-a-720p and moon-2.0-gl-a-720p-fast, via /v1/videos. Both support text-only input, people references and up to 4 images + 3 videos + 1 audio file, 4–15 seconds and 16:9, 9:16, 1:1. Audio requires an accompanying image or video.
• Moon-2.5 GL: moon-2.5-gl-a, via /v1/videos, supports text-to-video, image/video/audio references, 480P / 720P, 4–30 seconds and 9:16, 16:9, 1:1, 21:9, 4:3, 3:4. Up to 30 images, 10 videos and 10 audio files are allowed, with at most 50 references total. Audio requires an image or video. Billing is per second.
• Moon-2 AC: moon-2.0-ac-a-480p, moon-2.0-ac-a-720p, moon-2.0-ac-b-720p-fast, moon-2.0-ac-c-720p, moon-2.0-ac-d-720p, moon-2.5-ac-a-480p, moon-2.5-ac-a-720p, moon-2.5-ac-b-720p and moon-2.5-ac-c-720p, all via /v1/videos. These are independent public models. Consult each model page and the pricing API for resolution, reference, duration and per-call/per-second billing rules.
• Moon-2 LD: moon-2.0-ld-a and moon-2.0-ld-b, via /v1/videos. Both have the same capabilities: 720P, 5–15 seconds, up to 9 images, 3 videos and 3 audio files, with at most 15 references total.
• Moon-2 AM 720P Fast: moon-2.0-am-720p-fast, entry /v1/videos, supports 720P, text-to-video, image-to-video, Photo/Video/Audience Reference and references[] Internet URL combination.
• Moon-2 IL 720P: moon-2.0-il-720p, via /v1/videos, supports 720P, 4–14 seconds, up to 9 images, 3 videos and 3 audio files, with at most 12 references total. Billing is per second.
• Moon-2 ME: moon-2.0-me-a-720p, moon-2.0-me-b-720p and moon-2.0-me-a-fast, via /v1/videos. All support 720P, 5 / 10 / 15 seconds and five aspect ratios. A/A Fast allow up to 9 images; B allows up to 9 images + 3 videos + 3 audio files, at most 15 total. Audio must accompany an image or video.
• Grok video: grok-video-1.5 and grok-image-video, via /v1/videos, with queries at /v1/videos/{task_id}. The current API requires 1 reference image for grok-video-1.5; grok-image-video allows up to 5 reference images.

Result Read

status must be checked before reading results. After success read URL or structured results in response; read fail_reason, error.code and error.message when failure.