Voidleap
agentic IDE, own harness, swap models mid-conversation
Last checked:
What is Voidleap?
Voidleap Code is a full agentic coding environment for developers building software with AI agents. It runs locally on your machine with no data routing through their servers, supporting 9 LLM providers (Anthropic, OpenAI, Gemini, GitHub Copilot, Ollama, LMStudio, OpenRouter, Azure, Bedrock) and local models. The tool provides live observability, customizable workflows, and privacy-first architecture.
Voidleap pricing
Pricing model: Freemium
Free (Voidleap Code 1.0 launched free for everyone as of 2026-08-16; no paid tiers mentioned)
Voidleap pros
- Local-first architecture with full privacy control—no code or prompts routed through their servers
- Multi-provider flexibility—switch between 9 providers per thread, agent, or turn, or run local models entirely
- Comprehensive observability—30+ charts tracking cost, tokens, latency, cache hits, and per-model/agent breakdown
- Granular context management—visually manage and prune context per round without restarting conversations
- Highly customizable—edit agents, skills, hooks, system prompts, and security rules at global, workspace, or project scope
Voidleap cons
- Desktop-only (Mac and Windows)—no web or mobile interface mentioned
- Requires license validation periodically; 24-hour offline limit with local models
- Setup complexity—requires bringing your own provider keys or managing local model infrastructure
Frequently asked questions about Voidleap
Where does my code go?
Decoupled inference with local-only architecture. All application data stays in ~/.voidleap/. You choose inference via cloud provider or local model; Voidleap Code never sees your code either way.
Can I work without a constant internet connection?
Yes, with local models your prompts and code stay on your machine. Voidleap Code only requires periodic license validation; once validated, you can stay offline for 24 hours.
Can I see what I'm spending?
Yes, 30+ charts cover cost trends per-model, per-agent, per-thread, per-tool, and cache hit rates, all exportable.
Which models can I use?
Nine providers: Anthropic, OpenAI, Gemini, GitHub Copilot, Ollama, LMStudio, OpenRouter, Azure, Bedrock. Switch per thread, agent, or turn with full tool-call parity for local models.
What execution modes are available?
Four modes: Default, Read-only, Careful (approve every action), and Yolo. Includes AST-aware command parsing and sensitive-path defaults (.ssh, .env*, .aws, .gnupg blocked by default).