Intelligence is inevitable. Trust is not.
nowara is a UK AI company. We map how AI is built, deployed and used, and test whether it can be trusted. That work powers AICAP today. Alongside it, we train, fine-tune and quantise models for specific tasks. We also research how AI systems are attacked, and how they're used in attacks, so we can defend against both.
products
Everything starts from the same map.
-
AICAP
Continuous AI assurance, compliance and security for systems that never stop changing. Knows what to test, generates the tests, and proves you did it. Includes a model arena for side-by-side procurement decisions.
-
Model Registry
A public, continuously updated record of AI models: releases, variants, quantisations and formats, with the provenance of every entry. Facts about models, not rankings of them.
-
Next
Our own models: task-specific foundation models, fine-tunes and quantisations, published with full lineage. Plus more AI security tooling from our red-team research. We'll list them here when they're ready to use.
model registry
Every model, every quant, every runtime.
A single LLM release fans out into dozens of downloadable artefacts. The registry tracks each one back to the model it came from, so you know exactly what you're running.
| model | publisher | variant | format | quant | params | size | released |
|---|---|---|---|---|---|---|---|
| Gemma 3 12B | it | GGUF | Q4_K_M | 12.2B | 7.3 GB | 2025-03-12 | |
| Mistral Small 3 | Mistral AI | Instruct 2501 | GGUF | Q4_K_M | 23.6B | 14.3 GB | 2025-01-30 |
| DeepSeek-R1-Distill | DeepSeek | Qwen 7B | safetensors | bf16 | 7.6B | 15.2 GB | 2025-01-20 |
| Phi-4 | Microsoft | base | GGUF | Q6_K | 14.7B | 12.0 GB | 2024-12-12 |
| Qwen2.5 7B | Alibaba | Instruct | AWQ | int4 | 7.6B | 5.6 GB | 2024-09-19 |
| Llama 3.1 8B | Meta | Instruct | GGUF | Q4_K_M | 8.0B | 4.9 GB | 2024-07-23 |
- Public data only
- Model cards, licences, weights metadata and release notes, each linked to its source.
- Always current
- New releases, re-uploads and community quants appear as they're published.
- No leaderboard
- The registry records what exists. Comparing and evaluating models is AICAP's job.
about
A UK company building AI tools for global problems.
Organisations now run models they didn't train, downloaded in forms their authors never published, and updated faster than anyone can review. We build across that landscape: models, AI assurance and AI security. Attacking AI is part of the work, because the only systems worth trusting are ones built by people who know how to break them.
Our mark is a sunflower head drawn from code. Each dot sits on the golden angle, the same rule a seed head uses to pack itself without gaps. It's how we like to work: many small, exact parts that add up to something you can see whole.