Ante is a self-contained coding agent that lives in your terminal and self-organizes. One ~15MB Rust binary, zero runtime dependencies, works with 12+ providers or fully offline with local GGUF models. Continuously evaled in public: #1 same-model agent on Terminal-Bench 2.1, every result pinned to a build you can download and audit.
- Primary: https://antigma.ai
- Docs: https://docs.antigma.ai
- Install / landing: https://ante.run
- GitHub: https://github.com/AntigmaLabs/ante-preview
Free. The ante binary is free during the alpha preview, commercial use included (Binary Preview Terms). Docs, protocol, SDK, and eval pipeline are open source under Apache 2.0. Best-fit PH option: Free or Payment not required; on sites with an open-source flag, mark open source.
Advantages:
- Continuously evaled and evolved, in public: #1 same-model agent on Terminal-Bench 2.1; every result pins a downloadable build and links the raw Harbor run for audit (live at https://antigma.ai/eval). With open-weight GLM 5.2 Ante scores 74.6%, a top-7 slot on the public verified leaderboard.
- A fraction of the footprint: ~7× less peak memory, ~9× less average CPU, ~5× less disk I/O than Claude Code across the same 20 parallel tasks.
- Natively offline: built-in inference engine runs any GGUF model with no key, account, or internet.
- Zero vendor lock-in: one ~15MB self-contained binary, 12+ providers, switch freely, no account required.
Social Links:
- X (Twitter): https://x.com/antigma_labs
- Discord: https://discord.gg/CbAsUR434B
- GitHub: https://github.com/AntigmaLabs
- Hugging Face: https://huggingface.co/Antigma
- LinkedIn: https://www.linkedin.com/company/antigma-labs
Top comments (0)