AI & Agents
nanointerpret
Last updated 2026-08-27 · benchmark measured 2026-08-27 — deterministic & reproducible
Interactive tool for exploring and intervening on LLM sparse autoencoder features.
Is nanointerpret production-ready?
Legit.Show scores nanointerpret 61 out of 100 — the simple average of its 7 measured frames. Legit.Show ran its deterministic 7-Frame production-readiness benchmark on nanointerpret (public-surface assessment), measured from the public surface with no LLM in the scoring path. Its strongest frame is Accessibility; its weakest is Discoverability. Every frame it averages is published with its evidence on the Legit.Show listing.
The 7 Frames
- Performance — 76/100
- Accessibility — 100/100
- Security — 45/100
- Privacy — 25/100
- Reliability — 67/100
- Standards — 92/100
- Discoverability — 25/100
What we measured
- Security headers present: X-Content-Type-Options, Referrer-Policy.
- No Content-Security-Policy and no HSTS.
- Served over HTTPS with a valid certificate.
- Real Lighthouse performance run — 142 ms to first byte.
- 0 of 0 sampled routes reachable.
- No privacy policy found.
- Sets cookies / loads scripts with no consent prompt.
- Discoverable: sitemap.
Who it's for
AI researchers · ML engineers · LLM interpretability enthusiasts · neuroscientists studying neural networks
Visit nanointerpret → · Alternatives to nanointerpret → · How this was measured →