BestLLMfor Your hardware. Your LLM. Your call.
The Local Copilot Kit APIOpen data Find my LLM
Model fiche

OSINT Researcher 4B

By Yemen-JPT · YE

Updated 2026-08-28

chat small multilingual
Parameters
4B
License
Apache 2.0
Context
32k
VRAM (Q4)
2.3 GB
Released
May 2026

Overview

A 4B Qwen3 fine-tune from Yemen-JPT specialized for open-source-intelligence research and investigative journalism, with native Arabic/English bilingual support.

When to pick this model

  • OSINT workflows requiring bilingual Arabic/English processing
  • Investigative journalism research assistance
  • Lightweight local deployment on modest hardware
  • Niche use cases where a general model underperforms on Arabic-language sourcing

VRAM requirements by quantization

VRAM REQUIRED (GB)Q4_K_M2.3 GBQ5_K_M2.8 GBQ8_04.3 GBFP168 GB
QuantizationVRAM required
Q4_K_M (recommended)2.3 GB
Q5_K_M2.8 GB
Q8_04.3 GB
FP16 (no quantization)8 GB

VRAM figures include model weights plus a typical 8k KV cache and ~600 MB runtime overhead (Ollama / llama.cpp baseline). Add headroom for higher context lengths.

In practice, OSINT Researcher 4B fits an 8 GB consumer card at Q4_K_M (2.3 GB). Stepping up to Q8_0 nearly doubles the footprint to 4.3 GB, and unquantized FP16 weights take 8 GB — plan your GPU around the Q4 or Q5 figure unless you specifically need the higher fidelity.

Without a GPU, OSINT Researcher 4B needs roughly 5 GB of system RAM to run on CPU via llama.cpp or Ollama — workable for background jobs, but far slower than GPU inference. Throughput estimates from our compatibility engine: around 32 tokens/sec on entry-level GPUs, on the order of 50 tokens/sec on a mid-range card, and up to 75 tokens/sec on high-end hardware — assuming the chosen quantization fully fits in VRAM.

What hardware do you need

The table below matches OSINT Researcher 4B to common GPU memory tiers, using the highest-fidelity quantization that fully fits each card class. Spilling layers to system RAM works but costs most of the speed, so size your card to the quantization you actually want to run.

GPU memoryExample cardsBest fit for OSINT Researcher 4B
8 GBRTX 5070 Laptop, RTX 5060, RTX 5060 Ti 8GBFP16 (8 GB used)
12 GBRTX 5070, RTX 5070 Ti Laptop, RTX 4080 LaptopFP16 (8 GB used)
16 GBRTX 5080, RTX 4080 Super, Radeon RX 9070 XTFP16 (8 GB used)
24 GBRTX 4090, Radeon RX 7900 XTX, RTX 5090 LaptopFP16 (8 GB used)
32 GBRTX 5090FP16 (8 GB used)

Which GPU should you buy to run OSINT Researcher 4B?

To run OSINT Researcher 4B locally at Q4, you need ~2.3 GB of VRAM. The best value for this is a RTX 5060 (8 GB VRAM).

Check RTX 5060 price on Amazon →Check RTX 5060 price on Newegg →

As an Amazon Associate, BestLLMfor earns from qualifying purchases, at no extra cost to you. It does not influence our independent rankings.

Strengths

  • Fully open Apache 2.0 license
  • Native Arabic/English bilingual capability
  • Compact footprint (~2.3GB VRAM at Q4)
  • Purpose-built for OSINT and journalism workflows

Limitations

  • Niche specialization, not a general-purpose model
  • Limited reasoning capacity at 4B scale
  • Few public benchmarks to validate performance claims

Typical workloads

In our catalog grid, OSINT Researcher 4B is filed under OSINT Research, Arabic-English Chat, Journalism — the use cases where its size/quality trade-off makes the most sense. Its tags translate to concrete workloads: multilingual workloads.

The 32k-token context window covers long chats and mid-sized documents, though very large retrieval workloads will need chunking. The Apache 2.0 license is permissive, so shipping it inside a commercial product raises no special legal questions.

Architecture & training

Architecture: Dense transformer · 4B parameters (Qwen3 base) · OSINT/journalism fine-tuning

Training: Yemen-JPT fine-tune on Qwen3-4B. Specialized for open-source research, investigative journalism, and bilingual Arabic/English processing.

Verdict

A compact, bilingual niche tool for OSINT and journalism work — not a general assistant.

Quick start

# HuggingFace : Yemen-JPT/OSINT-Researcher

Or use the open-source MCP server to query this model from Claude Desktop, Cursor, or any MCP-compatible client.

Similar models worth comparing

Frequently asked questions

How much VRAM does OSINT Researcher 4B need?

At the recommended Q4_K_M quantization, OSINT Researcher 4B needs about 2.3 GB of VRAM. Q8_0 takes 4.3 GB, and unquantized FP16 weights take 8 GB.

Can OSINT Researcher 4B run without a GPU?

Yes — with roughly 5 GB of system RAM it runs CPU-only through llama.cpp or Ollama. Expect a fraction of GPU speed, which is fine for background or batch jobs but slow for interactive chat.

What context window does OSINT Researcher 4B support?

OSINT Researcher 4B supports a 32k-token context window (32,768 tokens).

Can I use OSINT Researcher 4B commercially?

Yes. OSINT Researcher 4B is released under Apache 2.0, a permissive open-source license that allows commercial use, modification and redistribution.

How fast is OSINT Researcher 4B on consumer hardware?

Our compatibility engine estimates on the order of 50 tokens/sec on a mid-range GPU and up to 75 tokens/sec on high-end cards, assuming the quantization fully fits in VRAM.

Which quantization of OSINT Researcher 4B should I download first?

Start with Q4_K_M (2.3 GB) — the standard size/quality sweet spot. Step up to Q5_K_M or Q8_0 only if you have VRAM headroom. It already fits an 8 GB card at FP16.

Tools

Is OSINT Researcher 4B the right pick for you?

Compute self-hosted ROI → Back to catalog