iQuantize: open-source desktop workspace for guided LLM quantization to GGUF, MLX and ONNX
Zipidyzip · reddit · 2026-11-06
A Reddit user released iQuantize (Apache 2.0), an end-to-end desktop quantization workspace for LLMs.
- Load local Safetensors or pick a model from Hugging Face
- Step-by-step guided conversion to GGUF, MLX, and ONNX with live visualization of per-tensor quantization and precision mix
- Compare results against the original and publish back to Hugging Face with a full model card
Mostly tested on Apple Silicon so far; Windows/Linux builds exist but are untested in depth and likely buggy. Contributors welcome. Repo: github.com/Fazmin/iQuantize
More from Infra
- Fudan's D3D decouples diffusion drafting for asynchronous speculative decoding speedups — Fudan-University · 2026-11-06
- ThursdAI pod: midterms, 40 datacenter ballot measures, $3T wipeout — is AI over? — altryne · 2026-11-06
- sshhole Bug Analysis: Alpine/MUSL Systems Hit Hard, ~100-500M OpenWRT Routers Were at Risk — esrtweet · 2026-11-06
- Smarter Compact: MCP server trims agent context, 72% cheaper long sessions with full recall — goost95 · 2026-11-06
- TensorFold 1.2.3 boosts Flash Next decode 21–23%; Apache-2.0 Ripple-65B Preview lands — AIFlow_ML · 2026-11-06
- "What am I supposed to do if the data center bubble pops" — AI infra circle jokes — max_paperclips · 2026-11-06