Introducing the Horus 1.0 Models
A state-of-the-art family of Large Language Models, engineered for deep reasoning, multilingual precision, and efficient on-device intelligence.
Models & Architecture
Horus 1.0 4B
Text-to-Text
Horus 1.0 4B is optimized for high-performance extraction and generation. Built on a transformer-based architecture, it balances computational efficiency with exceptional knowledge density.
100%
Open Source Weights
Global Community Impact
Total Horus Models Downloads
Trusted by researchers and developers worldwide across Hugging Face and enterprise mirrors.
How to Use
Install the framework in Python
pip install neuralnode⚠️ Required Dependencies
You must also install these dependencies for GGUF models to work: llama-cpp-python, huggingface_hub
(Not needed: torch, transformers, accelerate, bitsandbytes)
pip install neuralnode llama-cpp-python huggingface_hubImplementation Guide
Basic setup and loading:
import neuralnode as nn
# Choose your model version (replace with your preferred version)
MODEL_ID = "tokenaii/Hours-1.0-4B-GGUF/Horus-1.0-4B-Q6_K.gguf" # 4-bit for low VRAM
# Download and load
model = nn.HorusModel(MODEL_ID).load()
# Use immediately
response = model.chat([{"role": "user", "content": "Hello!"}])
print(response.content)One-liner: chat in a single chain:
import neuralnode as nn
# One-liner: create model, load it, and chat in a single chain
response = nn.HorusModel("tokenaii/Hours-1.0-4B-GGUF/Horus-1.0-4B-Q6_K.gguf").load().chat(
[{"role": "user", "content": "What is AI?"}]
)
# Print the response content
print(response.content)Model Configurator
Customize your setup and get a ready-to-run code snippet
Select Quantized Version
Browse All Voices
List 20 Replica voices programmatically.
import neuralnode as nn
# Get the full list of all available Replica TTS voices
voices = nn.replica_voice_list()
# Print all voices with their IDs and languages
for voice in voices:
print(voice)List Available Models
See the full Horus LLM series catalog.
import neuralnode as nn
# List available Horus models
nn.Horus.model.print()Performance Metrics
| Benchmark Metric | Horus 1.0 (4B) | Llama-3.1 (8B) | Avg Peers |
|---|---|---|---|
| Reasoning (MMLU) | 85.0% | 69.0% | 71.0% |
| Coding (SWE-bench) | 55.0% | 30.0% | 40.0% |
| Terminal (Logic) | 84.0% | 55.0% | 50.0% |
| HellaSwag (Knowledge) | 85.0% | 80.0% | 75.0% |
| Browse (General) | 88.0% | 45.0% | 40.0% |
Hardware Capability
Horus 1.0 4B
Minimum: 16GB RAM / 8GB VRAM
Target: Int4 / GGUF
Quantized Versions (GGUF)
| Format | File Size | Min RAM | Min VRAM | Quality | Best For |
|---|---|---|---|---|---|
| F16 | 9.03 GB | 12 GB | 10 GB | Maximum quality | High-end GPUs (RTX 3090, A100) |
| Q8_0 | 4.8 GB | 6 GB | 5 GB | Near-lossless | RTX 3060 12GB, RTX 4060 |
| Q6_K | 3.71 GB | 5 GB | 4 GB | Excellent | RTX 3060, RTX 4060 Laptop |
| Q5_K_M | 3.23 GB | 4 GB | 3.5 GB | Very Good | GTX 1650, RTX 3050 |
| Q4_K_M | 2.78 GB | 3.5 GB | 3 GB | Good | Entry-level GPUs, CPU-only |
Standardized Verification
Join the Horus Development Team
Contribute to the evolution of open-source intelligence. We are seeking researchers and engineers dedicated to pushing the architectural limits of LLMs.