Saltar al contenido
iCrisol

Real project status

The first Crisol already breathes.

This is not a paper. The first Crisol Mini has been trained from scratch: 5,000 steps, without a single NaN or Inf, the loss falling from 45 to a best of 3.5 and causal reasoning switching on by itself. All in 24 hours and for ten cents of electricity. Here is the real status — no makeup, including what is still missing.

~2.13 B
parameters trained from scratch
0
NaN / Inf in 5,000 steps
45 → 3.5
loss (CE): PPL 64,000 → 34
0.0087
epochs seen · 164 M tokens

A model trained from scratch, not a borrowed fine-tune.

The first Crisol Mini — ~2.13 B parameters, a single expert per layer — was forged entirely on its own architecture: 12 layers, 5 universal slots, an O(1)-cost HoloBinder, holo_dim 4096, a NAR of 2048 axes and a NOE of 2048 dimensions. Not a single weight inherited from an external model.

Across 5,000 steps it produced not a single NaN or Inf: bounded gradients, healthy distributions, stable training from start to finish. And as the loss fell from 45 to a best of 3.5, reasoning woke up on its own — causal confidence rose from 0 to ~0.8 and the geometry aligned with the NAR axes. It doesn't just predict: it begins to reason.

And its whole world already lives in its 64,000-piece vocabulary — the names of its agents, its validators and its memory structures:

HistorianAgentZ NOEValidator HoloBinder CoreEncoder MemoryExpert CausalStore NomotheticZ InquisitorZ ArchitectZ SuenoCausal IdentidadStore NAR axes

Phased roadmap

From clean ground to an organism that breathes.

Six phases done, one in progress, one pending. The launch target is September 2026.

  1. Phase 0

    Preparation

    done

    Backups, a dedicated working branch, selective cherry-picks and Done criteria per sub-step. Clean ground before raising anything.

  2. Phase 1

    Five new components

    done

    CoreEncoder, MemoryExpert, IdentidadStore, CausalStore and SuenoCausalAutomatico. The pieces that turn a model into a persistent organism.

  3. Phase 2

    Backend integration

    done

    All five components stitched into the engine's stack, layers and enums. Not a script on top: part of the forward pass.

  4. Phase 3

    REST endpoint + Crisol Studio

    done

    Full control from the interface. Every parameter configurable, every process traceable, real-time metrics over WebSocket.

  5. Phase 4

    Pre-trainings

    done

    A 64,000-piece BPE tokenizer, a NAR of 2048 axiomatic axes and a NOE of 256 invariants. The substrate of reasoning and knowledge, before the corpus.

  6. Phase 5

    Forging the base model

    done

    The first Crisol Mini trained from scratch — ~2.13 B parameters, 5,000 steps without a single NaN. The loss fell from 45 to a best of 3.5 and causal reasoning switched on by itself. Forged after fixing the learning rate and all of Crisol Studio.

  7. Phase 6

    The five universal packages

    in progress

    Generating the .crisolpkg experts that travel through the Custodia cloud and import like apps. The container-ship starting to load.

  8. Phase 7

    E2E validation + launch

    pending

    End-to-end testing of the whole organism and public release. Target: September 2026.

Where Crisol does not yet compete.

Crisol is not a "GPT-killer". Saying otherwise would be dishonest. There are arenas where a frontier LLM, as of today, wins — and they deserve to be named plainly.

General fluency

A Crisol Mini does not write prose as polished or as broad as a frontier LLM trained on trillions of tokens. That is not its goal.

Encyclopedism

It does not try to know everything from memory. Knowledge arrives modularly, as .crisolpkg packages, not baked into the weights.

Multimodality

Today it reasons over and generates text. Vision, audio and other modalities are not part of v1.0.

Raw speed

A closed model served from a datacenter answers faster per request. Crisol trades milliseconds for sovereignty.

But it competes — and wins — where it truly matters.

Crisol is not playing to be one more, larger model. It is playing a different game.

Sovereignty 100% local, on your hardware, with no cloud and no mandatory telemetry.
Privacy Your data and your memory never leave the device.
Modularity Importable and removable experts, like apps.
Persistence Living memory across sessions — the conversation that lasts years.
Causality Real counterfactual reasoning, not mere correlation.

The why behind the status

The status is honest because the thesis deserves it.

If you want to understand what motivates each phase of this roadmap — why a sovereign organism and not a larger model — the manifesto explains it in full.

Read the manifesto