# Windflow: realtime AI video and world model API > Windflow runs realtime interactive AI video models as on-demand sessions. Teams plug them into their products and pay by the session second. Source: https://windflow.online/ · Windflow Labs ## Public model catalog # Windflow models > 10 hosted realtime video and world models on one session API, billed per ready second. Source: https://windflow.online/models · Windflow Labs ## Helios - API: https://windflow.online/models/helios - Implementation: production; live readiness is reported separately. - Architecture: 14B Diffusion Transformer, autoregressive chunked diffusion - Resolution: Negotiated per session - Playout: Negotiated per session - Chunk: 33 native frames in the production profile - Control latency: First chunk after roughly one chunk of generation - Inputs: Text prompt (required before start) and an optional reference image for image-to-video - Outputs: One continuous, steerable clip on the main_video track - Delivery: WebSocket (H.264) · WebRTC - Session protocol: sdk.realtime.v2 - Price: $0.0017 per ready second ## LongLive-2.0 - API: https://windflow.online/models/longlive-2 - Implementation: production; live readiness is reported separately. - Architecture: Autoregressive multi-shot DiT (Wan2.2-TI2V backbone) - Resolution: 1280 × 704 - Chunk: 29 frames - Scene budget: 48 chunks per scene; scene_cut opens a fresh budget - Inputs: Text shot prompts only — no reference image in this release - Outputs: A multi-shot clip on main_video; chunk_index resets per scene while session_chunk keeps counting - Delivery: WebSocket (H.264) · WebRTC - Session protocol: sdk.realtime.v2 - Price: $0.0021 per ready second ## SANA-Streaming - API: https://windflow.online/models/sana-streaming - Implementation: production; live readiness is reported separately. - Architecture: 2B streaming editing Diffusion Transformer - Resolution: 1280 × 704 native edit and delivery - Chunk: 24 frames - Control latency: A new edit prompt lands at the next chunk boundary - Inputs: A source clip you upload (≥ 41 frames) or a live webcam track, plus an edit prompt - Outputs: The edited stream on main_video, with original motion preserved - Delivery: WebSocket (H.264) · WebRTC - Session protocol: sdk.realtime.v2 - Price: $0.0019 per ready second ## JoyAI-Video-Edit - API: https://windflow.online/models/joyai-video-edit - Implementation: production; live readiness is reported separately. - Architecture: 16B multimodal DiT, chunk-autoregressive, two-step distilled - Resolution: 840 × 480 interactive tier; 1248 × 720 detail tier - Chunk: 8 frames - Control latency: About 0.9 s from camera frame to the worker socket at 840 × 480, before the browser cushion - Inputs: A source clip you upload or a live webcam track, an edit instruction, and optionally a reference image - Outputs: The edited stream on main_video, with source motion and identity preserved - Delivery: WebSocket (H.264) · WebRTC - Session protocol: sdk.realtime.v2 - Price: $0.0021 per ready second ## LingBot-World 2.0 - API: https://windflow.online/models/lingbot - Implementation: production; live readiness is reported separately. - Architecture: Seed-image conditioned navigable world model - Resolution: 1408 × 768 delivery (704 × 384 generation) - Chunk: Movement and look apply at the next generation-block boundary - Inputs: A seed image (uploadFile) and a text prompt, plus persistent WASD movement and look controls - Outputs: A navigable stream on main_video you steer in real time - Delivery: WebSocket (H.264) · WebRTC - Session protocol: sdk.realtime.v2 - Price: $0.0024 per ready second ## Matrix-Game 2.0 - API: https://windflow.online/models/matrix-game-2 - Implementation: production; live readiness is reported separately. - Architecture: Real-time AR world model (SkyReels-V2 base, Self-Forcing distilled, 3-step) - Resolution: 640 × 352 - Chunk: 12-frame AR blocks; actions land at the next block - Control latency: One block from input to pixels - Inputs: A seed image and persistent WASD/look controls (no text prompt) - Outputs: A navigable generated world stream on main_video - Delivery: WebSocket (H.264) · WebRTC - Session protocol: sdk.realtime.v2 - Price: $0.0024 per ready second ## DreamX-World-5B - API: https://windflow.online/models/dreamx-world-5b - Implementation: production; live readiness is reported separately. - Architecture: 5B controllable world generation model (Wan2.2-TI2V backbone) - Resolution: 704 × 1280 portrait in the production runtime; negotiated per session - Chunk: Action-sequence conditioned rollout - Inputs: A seed image, a text prompt, and movement/camera-control state - Outputs: A controllable world rollout on main_video - Delivery: WebSocket (H.264) · WebRTC - Session protocol: sdk.realtime.v2 - Price: $0.0023 per ready second ## ABot-World-0-5B - API: https://windflow.online/models/abot-world-0-5b - Implementation: production; live readiness is reported separately. - Architecture: 5B keyboard-conditioned world model (Wan2.2-TI2V backbone, LongForcing causal student) - Resolution: 1344 × 768 delivered (896 × 512 generation + learned super-resolution) - Chunk: 12-frame AR blocks; held keys are sampled at each block boundary - Inputs: A seed image, a text prompt, and persistent movement/camera key state - Outputs: A third-person navigable world stream on main_video - Delivery: WebSocket (H.264) - Session protocol: sdk.realtime.v2 - Price: $0.0023 per ready second ## MiniMax FastH3 - API: https://windflow.online/models/fast-h3 - Implementation: production; live readiness is reported separately. - Architecture: 35B MiniMax H3 text-to-video-and-audio model, DMD2-distilled to four DiT calls with 90% sparse VSA-H3 attention - Resolution: 960 × 544 - Playout: 24 - Chunk: 345 frames per window (14.38 s at 24 fps; published as the 15 s profile) - Control latency: A whole window. The pipeline materialises the clip before delivery, so time-to-first-frame equals one generation. - Inputs: A text prompt with visual, dialogue, soundscape, and music direction - Outputs: Video on main_video and generated audio downmixed to mono on main_audio, aligned per window - Delivery: WebRTC - Session protocol: sdk.realtime.v2 - Price: $0.0069 per ready second ## LTX-2.5 - API: https://windflow.online/models/ltx-2-5 - Implementation: production; live readiness is reported separately. - Architecture: 22B joint audio+video DiT, two-stage, generated one window at a time - Resolution: 832 × 448 measured; the canvas is configurable in multiples of 64 - Playout: 24 - Chunk: 241 frames per window (10.04 s) - Control latency: A whole window: the pipeline does not yield progressively, so time-to-first-frame equals one window's generation time. The reference still is published while it runs. - Inputs: A text prompt and an optional reference still - Outputs: Video on main_video and the audio generated with it on main_audio, aligned per window - Delivery: WebRTC - Session protocol: sdk.realtime.v2 - Price: $0.0040 per ready second ## Common questions ### What is Windflow? Windflow runs realtime interactive AI video models as on-demand sessions. Teams plug them into their products and pay by the session second. ### Which realtime video and world models can I run through the API? The public catalog includes Helios, LongLive-2.0, SANA-Streaming, JoyAI-Video-Edit, LingBot-World 2.0, Matrix-Game 2.0, DreamX-World-5B, ABot-World-0-5B, MiniMax FastH3, LTX-2.5. Choose by supported inputs, controls and outputs. A published runtime status describes implementation maturity; live capacity is reported separately. ### How is a realtime video API different from a video generation job? A realtime session keeps model state alive, delivers a video track and accepts supported controls while it runs. A video generation job takes an input and returns a completed result. Windflow provides sessions for live editing, continuous generation and navigable worlds. ### How much does the Windflow API cost? Public model rates range from $0.0017 to $0.0069 per ready session-second. Queueing and worker startup are free. A ready session remains billable when its video is paused or unwatched; release it to stop the meter. ### Can I embed an interactive world model in my game or app? Yes. Use a model that accepts movement and look controls, stream its generated video into your interface, and forward the user's controls during the session. Windflow offers JavaScript, React and Python integrations. Generated worlds have model-specific continuity limits and do not provide a deterministic physics engine. ### How can I try Windflow or bring my own model? Windflow is in private beta. Request access or a live demo; approved accounts receive $5 and 4 concurrent sessions. Model teams can submit licensed weights and a reproducible inference path through Bring a model. Guides: https://windflow.online/guides · Comparisons: https://windflow.online/compare · Use cases: https://windflow.online/use-cases