Chandra OCR team accuses interfaze-1-lite of copying their repo, worse scores

RexDouglass · x · 2026-10-08

The Chandra OCR team publicly accused the newly released open-weight interfaze-1-lite of being a copy: the repo allegedly reproduces Chandra OCR 2, renamed with a changed license, and performs worse — olmocr score 83.8 vs Chandra's 85.8.

Interfaze had pitched interfaze-1-lite as the first general open-weight model for deterministic work (OCR, speech-to-text, extraction, classification), returning confidence scores and bounding-box metadata on every run.

Original post →

More from Fun

• "Quietly" Is Such an LLM Word — The Telltale Sign of AI-Generated Copy — Salmaaboukarr · 2026-10-08
• ChatGPT App Controlling Claude Controlling ChatGPT: An Agent-Calling-Agent Moment — danshipper · 2026-10-08
• 1985 practitioner sides with Schmidhuber: backprop history isn't the US narrative — SchmidhuberAI · 2026-10-08
• Veteran artist-animator-coder welcomes AI tools that let anyone create and express themselves — dreamwieber · 2026-10-08
• The 'mathplosion' fable: AI Twitter's viral model-despair-in-CoT lore — repligate · 2026-10-08
• Dev shows off TypeScript compiler comparison: money and coding agents can't buy real craft — MarcJSchmidt · 2026-10-08

Fun channel →