TensorFold
Serves language models locally from a Zig binary on Apple Silicon and NVIDIA GPUs.
Not verified yetKnown vulnerabilities
How far pickfit got
pickfit hasn’t verified TensorFold yet, so there are no tested install steps.
What follows comes from its README only.
Use it
Before you install: known vulnerabilities
pickfit found 5 known high or critical vulnerabilities (6 findings, one per affected package version) in the packages a fresh install brings (OSV.dev, 3 Nov 2026). Weigh this before you install it.
npm:undici 7.30.0undici vulnerable to TLS certificate validation bypass via dropped tls options in Agent (GHSA-2m5v-pch6-74hx, high, fixed in 7.30.1)npm:zod 3.25.76Zod can exhaust memory while validating oversized constrained collections (GHSA-mrhx-c7xc-m3w5, high, fixed in 4.6.6)
See 4 more
npm:undici 7.30.0— GHSA-4w7w-5xrc-9h26 (high), fixed in 7.30.1npm:undici 7.30.0— GHSA-qrxm-fqhp-jwvm (high), fixed in 7.30.1npm:undici 7.30.0— GHSA-xjj6-f33q-929c (high), fixed in 7.30.1npm:zod 4.4.3— GHSA-mrhx-c7xc-m3w5 (high), fixed in 4.6.6
You need first
- A GPU is required
What it does
pickfit hasn’t run it yet, so these come from the README only.
- Serve language models as local serverNot tested
- Download checkpoints into Hugging Face cacheNot tested
- Run Metal or CUDA kernels directlyNot tested
- Teach running model new factsNot tested
Good for
- serve Qwen3.5-2B checkpoint as local-model
- download model checkpoints for offline serving
- send chat completion requests to local server