Mel - 2026 Pricing, Features, Reviews & Alternatives

Mel is a local-first agentic terminal for macOS, Windows, and Linux that unifies chat, shell, and code editing into a single desktop workflow. Developers and technical users describe tasks in plain English; a lead AI agent decomposes the work, spawns parallel worker agents or embedded CLI coding agents (claude, codex, opencode, and others) in split panes, polls progress, and integrates results. Every file operation — move, rename, write, delete — requires explicit user approval before execution, keeping the user in control throughout.

Mel ships with a built-in editor with LSP language support, a code review panel with diffs and inline comments, global search via ripgrep and git grep, and a project explorer with drag-to-prompt and attach-as-context capabilities. Conversations are saved as durable JSON, resumable with full model context, and automatically compacted. Project memory stores per-project notes server-side, recalled automatically at the start of each turn and shared across orchestration teams, so prior decisions carry forward across sessions without re-explanation. Auto model routing selects the best available model per turn across Anthropic, OpenAI, Gemini, Groq, DeepSeek, and Moonshot, with configurable bias modes (Responsive, Cost-efficient, Genius) and explicit per-conversation model pinning. Bring-your-own-key (BYOK) support lets the agent run on user-supplied provider keys at zero Mel credits.

Mel targets individual developers on the Free tier, power users on Pro and Max plans (up to 8 parallel orchestration agents and unlimited cross-session memory), and organizations on the Business plan with pooled credits and per-seat access. Additional capabilities include built-in web search with no separate API key, voice input, image and PDF attachments with vision routing, SSH and remote session support, and a command palette with autocomplete, splits, and tabs. The terminal is free with no account required.