Public experimental release49,295,872 parameters
quantum-1-pilot
Public legacy and base-model experiment for compact German completion research.
- Status
- Public experimental release
- Model
- German base-completion model
- Use
- German completion research
MODELS
Compact model experiments, documented clearly.
Explore two public experimental pilot releases and the source-linked, in-development quantum-1-echelon pipeline.
Model index
Filter models
Public experimental release49,295,872 parameters
quantum-1.6-pilot
A 49.3M-parameter experimental German completion model released after a reported continued-pretraining stage.
- Status
- Public experimental release
- Model
- German completion model
- Use
- F16 GGUF · llama.cpp
Pipeline and preflight stage — no trained modelParameter size not yet defined
quantum-1-echelon
The current strategic Quantum model line. Public evidence covers an architecture preflight, a validated tokenizer and Garden data-pipeline smoke tests; no Echelon model has been trained or released.
- Status
- No trained public model
- Model
- Tokenizer and data-pipeline evidence
- Use
- Echelon Base → later Chat stage
MODEL SYSTEM COMPARISON
Three model stages, compared on the same evidence scale.
Architecture, tokenizer and data-pipeline differences are shown separately so released artifacts, published validation and configured Echelon targets cannot be mistaken for the same kind of evidence.
ARCHITECTURE SCALE
From a 49.3M pilot to a configured 506.3M Echelon Base.
Parameter and context scale can be compared directly. Layer dimensions are shown only where a public source states them; the first pilot therefore keeps an explicit documentation gap.
Released artifact
quantum-1-pilot
- Parameters
- 49.3M released parameters
- Context
- 512 tokens
Detailed dimensions
Not published in pinned evidence
Detailed layer dimensions are not stated in the pinned public manifest.
Released artifact
quantum-1.6-pilot
- Parameters
- 49.3M released parameters
- Context
- 512 tokens
12 layers
- hidden size
- 512
- attention heads
- 8
- KV heads
- 8
The continued-pretraining release keeps the compact pilot architecture.
Configuration preflight only
quantum-1-echelon
- Parameters
- 506.3M configured parameters
- Context
- 2,048 tokens
26 layers
- hidden size
- 1,280
- attention heads
- 20
- KV heads
- 5
Proposed Echelon Base configuration; no trained weights or measured capability.