openrouter:google/gemma-4-31b-it:free
LOCAL KEEP AI
Home (/) · Documentation (/docs) · Free Models (/free-models) · Pricing (/pricing) · Downloads (/downloads) · FAQ (/faq) · About (/about) · Support (/support) · Privacy (/privacy) · Terms (/terms)
Local Keep AI Workspace — Private Local & Cloud AI Coding Assistant
The Local Keep AI interactive workspace provides a unified conversational coding, research, and document analysis environment powered by hundreds of local and cloud language models with zero telemetry surveillance.
Hybrid Local-First and Cloud Inference Architecture
Local Keep AI executes prompts either directly in your browser via WebGPU/WebLLM, on your local workstation via the lk CLI daemon (Ollama, llama.cpp, vLLM, MLX), or through zero-data-retention cloud endpoints. You can switch between free local models like Gemma 4, Qwen 2.5 Coder, and Llama 3.3 and frontier cloud models such as Claude Sonnet, GPT-4o, and DeepSeek R1 from the quick model bar at any time.
Multi-File Context, Code Execution & Document Analysis
Attach source files, screenshots, PDFs, or entire repository diffs directly into your conversation. Local Keep AI parses syntax trees, generates unified diffs, explains compiler diagnostics, and verifies patches against your test suite without training third-party models on your proprietary code.
Seamless Cross-Device Sync & Offline Lease Support
Conversations, custom system prompts, and workspace contexts synchronize across Web, macOS, Windows, Linux, iOS, and Android. When working offline on flights or air-gapped networks, Local Keep AI automatically routes inference to your downloaded on-device GGUF/WebGPU models.
Frequently Asked Questions
Are my chat prompts used to train external AI models?
No. Local Keep AI enforces strict zero-data-retention (ZDR) policies on cloud endpoints and runs local models 100% on your own hardware so your prompts and source code remain private.
How do I switch between free models and Pro cloud models?
Use the Quick Model Bar above the chat composer or open the full model selector to filter by local, free browser, or frontier cloud models.