- Entry date
- 20 January 2026
- Category
- Language
- Lead over the world
- a working baseline, not a lead
- Access
- 🔒 Protected
768 primitives instead of a token vocabulary — roughly a 10.8x compression on the same content, with a lexer and parser that actually run.
- 768
- active tokens/primitives
- ~10.8x
- compression vs. ordinary token streams
- 48
- lexer/parser tests
- 202,906
- training pairs, later 100% verified
Honest evaluation
The compression and tooling claim (768 primitives, ~10.8x, working lexer/parser) is proven; whether a model trained on this representation reasons better remains untested.
What we learned — a preview
The honest edges are open even when the engine is not. The full report — what it was, what we built, and the measured internals — is protected.
- A working lexer/parser and a verified training channel are real progress, but BrainLang is not itself a trained reasoning model.
- The gap between 'the DSL exists' and 'a model is trained on it end to end' was still open at the time of the audit.
- Compiling down to two lower-level targets (a GPU-native language and a JS embeddable form) adds real integration surface that wasn't fully exercised yet.
The deep body of “BrainLang — a language for thought, not text, compiled to two lower targets” is behind access.
We open the demos, the specs, and the method; we protect the engines, the model internals, and anything that touches commercial, safety-critical work. This report describes an engine — so its details are gated, and no client, company, or project is named. Members can read it in full; if you have a genuine reason to see it, tell us who you are and why.