flux-3-video-edit

FLUX 3 Video Edit API: Python Wrapper for Black Forest Labs' Video Continuation (FLUX 3 Video Extend)

A Python wrapper for FLUX 3 Video Extend — Black Forest Labs' video-continuation mode within the FLUX 3 family, delivered via muapi.ai. FLUX 3 Video Extend continues an existing clip with new prompt-guided motion, scene development, and camera movement — with optional native synchronized audio — using Black Forest Labs' unified image/video/audio architecture, so the continuation stays physically grounded and consistent with the source clip.

Important: FLUX 3 does not ship a general instruction-driven video editor (no arbitrary re-grade, restyle, or object-swap on an existing clip). The live API is a continuation model — it extends a clip forward in time based on your prompt, it does not modify the clip's existing frames. This package's method names (edit_video, etc.) are kept for backwards compatibility, but they call the real flux-3-video-extend endpoint. Prefer the new extend_video() alias in new code.

Status: Live. flux-3-video-extend and its fast/cheap draft sibling flux-3-video-extend-draft are both live on MuAPI today.

Related Projects

• Flux-3-Dev-API — Python SDK for FLUX 3 Dev, plus the full FLUX 3 image/video family
• flux-3-video-upscaler — Python SDK for upscaling FLUX 3 (or any) video output
• flux-3-omni — Python SDK for FLUX 3's multi-reference Omni Reference mode
• flux-3-video-api — Python wrapper focused on FLUX 3 Text-to-Video and Image-to-Video
• awesome-flux-3-api-prompts — FLUX 3 API guide, prompt engineering, and a curated prompt library
• flux-3-comfyui — ComfyUI custom nodes for the FLUX 3 API
• Open-Generative-AI — open-source, self-hosted AI image & video generation studio (200+ models)

🚀 Why FLUX 3 Video Extend?

FLUX 3 is Black Forest Labs' newest frontier model — a unified multimodal system jointly trained across image, video, and audio, extendable to action prediction for robotics. FLUX 3 Video Extend brings that same architecture to continuing existing footage instead of only generating new clips from scratch:

• Prompt-Guided Continuation: A single text prompt drives what happens next in the clip — no masks, keyframes, or timeline tools required.
• Physically Grounded: The continuation stays consistent with the clip's original lighting, motion, and geometry rather than warping the scene.
• Native Synchronized Audio: Optional audio generation for the extended portion, at no extra charge.
• Shares FLUX 3's Architecture: The same unified model that powers FLUX 3 Video and FLUX 3 Action, so quality and behavior track the flagship model.
• Developer-First: Simple Python SDK on top of MuAPI's unified infrastructure — no separate account or waitlist needed once you have a MuAPI key.

🌟 Key Features

• ✅ FLUX 3 Video Extend: Continue an existing video clip with new prompt-guided motion via extend_video().
• ✅ FLUX 3 Video Extend Draft: Fast, lower-cost draft mode for testing continuity before a final-quality extension via extend_video_draft().
• ✅ File Upload: Upload local videos directly using upload_file().
• ✅ MCP Server: Use FLUX 3 Video Extend as a Model Context Protocol server for Claude Desktop, Cursor, and other MCP clients.

---

🛠 Installation

Via Pip (Recommended)

pip install flux-3-video-edit-api

From Source

git clone https://github.com/Anil-matcha/flux-3-video-edit.git
cd flux-3-video-edit
pip install -r requirements.txt

Configuration

Create a .env file in the root directory and add your MuAPI API key:

MUAPI_API_KEY=your_muapi_api_key_here

---

🤖 FLUX 3 Video Extend MCP Server

Use FLUX 3 Video Extend as an MCP (Model Context Protocol) server so AI clients (like Claude Desktop or Cursor) can directly invoke the video-continuation tool.

Running the MCP Server

1. Ensure MUAPI_API_KEY is set in your environment.
2. Run the server:

   python3 mcp_server.py
   

3. To test with the MCP Inspector:

   npx -y @modelcontextprotocol/inspector python3 mcp_server.py
   

---

💻 Quick Start (Python)

from flux3_video_edit_api import Flux3VideoEditAPI

# Initialize the client
api = Flux3VideoEditAPI()

# Continue an existing video clip
print("Submitting FLUX 3 Video Extend task...")
submission = api.extend_video(
    prompt="The camera pulls back to reveal a sunrise breaking over the skyline",
    video_url="https://example.com/clip.mp4",
    resolution="1080p",
    duration=5,
)

# Wait for completion
result = api.wait_for_completion(submission["request_id"])
print(f"Success! Output: {result['outputs'][0]}")

---

📡 API Endpoint & Reference

FLUX 3 Video Extend

Endpoint: POST https://api.muapi.ai/api/v1/flux-3-video-extend

curl --location --request POST "https://api.muapi.ai/api/v1/flux-3-video-extend" \
  --header "Content-Type: application/json" \
  --header "x-api-key: YOUR_API_KEY" \
  --data-raw '{
      "prompt": "The camera pulls back to reveal a sunrise breaking over the skyline",
      "video_url": "https://example.com/clip.mp4",
      "resolution": "1080p",
      "duration": 5,
      "generate_audio": true
  }'

Python SDK:

submission = api.extend_video(
    prompt="The camera pulls back to reveal a sunrise breaking over the skyline",
    video_url="https://example.com/clip.mp4",
    resolution="1080p",
    duration=5,
)
result = api.wait_for_completion(submission["request_id"])
print(result["outputs"][0])

FLUX 3 Video Extend Draft (fast, lower-cost)

Endpoint: POST https://api.muapi.ai/api/v1/flux-3-video-extend-draft

submission = api.extend_video_draft(
    prompt="The camera pulls back to reveal a sunrise breaking over the skyline",
    video_url="https://example.com/clip.mp4",
    duration=5,
)

The draft endpoint has no resolution parameter — it's tuned purely for speed and cost during iteration.

---

📖 Method Reference

Method	Parameters	Description
extend_video (alias: edit_video)	prompt, video_url, resolution, aspect_ratio, duration, generate_audio	Continue an existing video clip with new prompt-guided motion.
extend_video_draft	prompt, video_url, aspect_ratio, duration, generate_audio	Fast, lower-cost draft mode of the same continuation.
upload_file	file_path	Upload a local file (image or video) to MuAPI for use in generation tasks.
get_result	request_id	Check task status for a FLUX 3 Video Extend generation.
wait_for_completion	request_id, poll_interval, timeout	Blocking helper that polls until the task completes.

---

💰 Pricing

Variant	Price
Video Extend, 720p	$0.25/second
Video Extend, 1080p	$0.42/second
Video Extend Draft	$0.09/second (flat)

Billed duration rounds up to the next whole second, capped at 20 seconds. Native audio generation is included at no extra charge.

---

🔗 Official Resources

• FLUX 3 Announcement (Black Forest Labs): bfl.ai/models/flux-3
• Playground — FLUX 3 Video Extend: muapi.ai/playground/flux-3-video-extend
• Playground — FLUX 3 (all variants): muapi.ai/flux-3
• Playground — FLUX Kontext (available today): muapi.ai/flux-kontext
• API Provider: MuAPI.ai

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

---

Keywords: FLUX 3 API, FLUX 3 Video Extend, FLUX 3 Video Extend API, Black Forest Labs FLUX 3, FLUX 3 Python SDK, FLUX 3 video continuation, AI video generation API, MuAPI, Python video generation SDK.