- Entry date
- 20 June 2026
- Category
- Language
- Lead over the world
- the thesis, not the lead
- Access
- 🔒 Protected
A language an AI writes, and a machine proves — the trust isn't in the code, it's in the agreement between two independent compilers.
- NovaLang → QUANTA
- the codename that became a standalone project
- self-hosting
- the core design thesis: lexes and parses itself, in itself
- 20 Jun 2026
- the dated point this entry marks in the arc
Honest evaluation
The self-hosting/dual-compiler mechanism is real and functioning; full self-hosting and full gate coverage are both still incomplete, per the linked research page.
Runnable proof — see it work
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.
- Still short one or two language features to fully self-host, per the current research-page state — this entry doesn't claim that gap is closed.
- The honesty census (570 chapters audited, ~59% confirmed genuinely real, 60 quarantined, per the research page) is the current bound on how much the gate can vouch for — a real, open limit, not a footnote.
- Compiler and codegen internals are deliberately kept off this page as core IP; the measured specifics live on the linked research pages.
The deep body of “QUANTA — from a GPU-native-language codename to a language that proves itself” 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.
Proofs & sparks
We demonstrate rather than assert. Each ✅ proof is a visible result with a hard figure.
- qc2 self-hosts, byte-identical53 / 53 byte-identicalthe compiler recompiles itself byte-for-byte under diverse-double-compilation (gcc ≡ clang), 53/53 — self-hosting you can check bit-for-bit.
- Register-allocation solver4.2× faster · 3× smallerthe register allocator drops from 25.8 to 9.6 cycles/call — 4.2× faster and 3× smaller code — and the cost model picks the strategy, reaching 0.8 TFLOP/s with no BLAS.
- Certifiable by constructiona memory leak is a compile errorqc2 --safety nasa-p10.safety rejects six classes of violation (no heap after init, bounded loops, no recursion, function length, assert density, checked returns) and issues a certificate on clean code.