Generate
Prompt
⚡

Nothing generated yet

Pollinations.ai works free with no API key.
Add keys in API Keys for more models.

Provider & Model
Output Format
🖥 YouTube
📱 Shorts
🎵 TikTok
📷 Instagram
✨ 4K
⬜ Square
Parameters
90
0.7
40
Storyboard
Build scene-by-scene — click a scene to edit it
+ Add Scene
🎬
Select a scene to preview
Select a scene to edit
Characters
Upload reference images — maintain consistency across all generations
🎯 Character Consistency: Upload a reference photo or generate a character. Once set as Active, all image/video generations will attempt to maintain their appearance. Reference images are injected into the prompt automatically.
+ Add Character Upload reference image
Agent Pipeline
Plan → Generate → Critique → Refine — with full shared context
🧠 All agents share a unified context object. Every stage receives all prior outputs. Your Forge Folder files (claude.md, agents.md, instructions.md) are automatically injected as system prompts — agents always behave per your rules.
Pipeline Stages idle
Brief / Input
Live Log
// Ready to run
Shared Context
No context yet
ACTIVE MODELS:
Multi-AI Chat — all active models respond with full memory + workspace context
Memory
200
Response Mode
Quick Actions
Model Hub
Enable models — workspace files are injected as system prompt for every model
💡 Toggle models on/off. Active models appear in Multi-AI Chat and Agent Pipeline. All models read your Forge Folder before responding — enabling fully consistent, instruction-governed AI behavior.
Forge Folder
All AI agents read these files before every action
📁 How it works: Before every API call, FORGE reads all workspace files and injects them into the agent's system prompt. claude.md = global behavior. agents.md = agent roles. skills.md = capabilities. instructions.md = rules. Agents always obey what you write here.
📁 /forge
Select a file
No file open 0 chars
API Keys
Stored in IndexedDB on your device — never sent anywhere except directly to the provider
🔒 All keys stored locally in IndexedDB. No server. No account. No data leaves your device except direct API calls to providers you configure. To get Groq & NVIDIA NIM keys for free, see the links below.
Add Custom / Unlimited APIs
Works with Ollama (local), LM Studio, OpenRouter, Together AI, or any OpenAI-compatible endpoint.
Settings
Workspace
All workspace files stored in IndexedDB → forge://workspace. Files persist between sessions.
Data & Privacy
Everything stored locally. No account needed. No telemetry. Direct API calls only.
Deploy — Cloudflare Pages (Free)
1. Go to pages.cloudflare.com
2. Create project → Direct Upload
3. Upload forge.html
4. Free global CDN hosting, forever
Open Cloudflare Pages →
About FORGE
FORGE v3.0
AI Creative Studio
Luma AI · Groq · NVIDIA NIM · Gemma 4
OpenAI · Anthropic · Google · HuggingFace
fal.ai · ElevenLabs · Pollinations
+ Any OpenAI-compatible endpoint
Single HTML file. Works offline. Open in any browser.