CelesteImperia/Llama-3.1-8B-Instruct-Platinum-GGUF

Llama-3.1-8B-Instruct-GGUF (Platinum Series)

This repository contains the Platinum Series universal GGUF release of Llama-3.1-8B-Instruct. This collection represents a monumental leap in local AI, featuring a 128k context window and state-of-the-art reasoning capabilities distilled from the 405B model. Optimized for enterprise-grade financial and regulatory analysis.

📦 Available Files & Quantization Details

File Name	Quantization	Size	Accuracy	Recommended For
Llama-3.1-8B-Instruct-Platinum-F16.gguf	FP16	~16.1 GB	100%	Master Reference / Benchmarking
Llama-3.1-8B-Instruct-Platinum-Q8_0.gguf	Q8_0	~8.1 GB	99.9%	Platinum Reference / High-Fidelity
Llama-3.1-8B-Instruct-Platinum-Q6_K.gguf	Q6_K	~6.6 GB	99.8%	Complex Regulatory Reasoning
Llama-3.1-8B-Instruct-Platinum-Q5_K_M.gguf	Q5_K_M	~5.8 GB	99.5%	Balanced High-End Performance
Llama-3.1-8B-Instruct-Platinum-Q4_K_M.gguf	Q4_K_M	~5.0 GB	99.0%	Default / General Purpose

---

🐍 Python Inference (llama-cpp-python)

To run these engines using Python:

from llama_cpp import Llama

llm = Llama(
    model_path="Llama-3.1-8B-Instruct-Platinum-Q8_0.gguf",
    n_gpu_layers=-1, # Target all layers to NVIDIA/Apple GPU
    n_ctx=32768      # Supports up to 131,072 context tokens
)

output = llm("Analyze the 2026 RBI Internal Ombudsman Directions.", max_tokens=512)
print(output["choices"][0]["text"])

---

💻 For C# / .NET Users (LLamaSharp)

This collection is fully compatible with .NET applications via the LLamaSharp library.

using LLama.Common;
using LLama;

var parameters = new ModelParams("Llama-3.1-8B-Instruct-Platinum-Q8_0.gguf") {
    ContextSize = 8192,
    GpuLayerCount = 33 
};

using var model = LLamaWeights.LoadFromFile(parameters);
using var context = model.CreateContext(parameters);
var executor = new InteractiveExecutor(context);

Console.WriteLine("Platinum Finance Engine Active.");

---

🏗️ Technical Details

• Optimization Tool: llama-quantize (CMake Build 2026-03-26)
• Architecture: Llama 3.1 (8B)
• Hardware Validation: Dual-GPU (RTX 3090 + RTX A4000)

---

☕ Support the Forge

Maintaining the production line for high-fidelity models requires significant hardware resources. If these tools power your research or industrial projects, please consider supporting the development:

Platform	Support Link
Global & India	Support via Razorpay

Scan to support via UPI (India Only):

---

Connect with the architect: Abhishek Jaiswal on LinkedIn