Family Qwen3 · 4B parameters

OSINT Researcher 4B

OSINT Researcher (4B Qwen3 fine-tuned on Yemen-JPT): a bilingual Arabic/English assistant for journalism and open-source research. Released May 2026.

Yemen-JPT·License Apache 2.0·Context 32k tokens·Output May 2026·Fits within the 128 GB of the GIGABYTE AI TOP ATOM← Catalog

01What it can do

Strengths
  • Completely free Apache 2.0
  • Native Arabic / English bilingual
  • Compact (2.3 GB VRAM Q4)
  • Specialized in OSINT and journalism
Limitations to know
  • —Niche model (not general-purpose)
  • —Capabilities limited by 4B size
  • —Few public benchmarks
Architecture
Dense transformer · 4B parameters (Qwen3 base) · OSINT/journalism fine-tuning
Training
Fine-tune Yemen-JPT on Qwen3-4B. Specialization in open-source research, investigative journalism, and Arabic/English bilingual processing.
Ideal for
OSINT / journalismArabic-English bilingualOpen-source research

04Install

Install Ollama for your OS. Check the model and its quantization before downloading. Start with 4096 tokens of context, then check placement with ollama ps. A command below is not proof that a test was run on your machine.

$# HuggingFace : Yemen-JPT/OSINT-Researcher
⚠
First download: between 2 and 40 GB depending on the selected quantization. Plan for sufficient disk space; a stable connection is recommended. Subsequent launches are instant.

02Required memory

Approximate GPU VRAM required to run this model, including 4k tokens of context overhead. For a longer context, add ~1 GB per 8k-token increment.

Q4_K_M
The lightest, ~5% loss
2.3 GB
Q5_K_M
Good quality/size compromise
2.8 GB
Q8_0
Nearly indistinguishable from FP16
4.3 GB
FP16
Full precision — server use
8 GB
Fallback CPU · If you don't have a GPU, allow 5 GB of RAM minimum to run this model at reduced speed.

What hardware do you need for OSINT Researcher 4B?

To run OSINT Researcher 4B locally with Q4 quantization, you need about 2.3 GB of VRAM. An option to compare: RTX 5060 Ti 16GB (ASUS Prime) — leave some headroom for the system and context; check engine compatibility with the GPU.

Current offer: RTX 5060 Ti 16GB (ASUS Prime)
AmazonSee price →

Affiliate links — commission possible at no extra cost to you; independent recommendations. As an Amazon Associate, BestLLMfor earns from qualifying purchases.

On the go: OSINT Researcher 4B also runs on a RTX laptop PC (16 GB of VRAM) →

This model in your private ChatGPT, without the cloud

Your private ChatGPT, free, on your own machine in an hour — LM Studio, Ollama, Open WebUI, your documents, no cloud.

  • Lifetime online access
  • PDF + files
  • Lifetime updates

03Expected speed

Tokens generated per second in Q4_K_M, 4k context. Beyond 20 t/s, reading is comfortable. Below 10 t/s, that's just for testing.

Entry-level
~32t/s
GTX 1650, RX 6600, MBA M2 8GB
Mid-range
~50t/s
RTX 4060, 4070, MBP M3 Pro
High-end
~75t/s
RTX 4090, M4 Max, Radeon 7900