TurinTech

Smaller GGUF versions of open models, each tested against its original and run on stock llama.cpp. TurinTech builds Artemis, a code optimisation platform; the decision models and Gemma files here were compressed with techniques it discovered.

Download decision models

Artemis Edge: Decision Models: Clef, Clef-Flash, Perplexity's Decider, Kev-9B and Kev-4B. They return a probability for each allowed answer through llama.cpp's /v1/systemone endpoint and are not chat models. Start with Clef-Flash (4.9 GB), the smaller Clef.

Download compressed chat models

Artemis Edge: Compressed LLMs: Gemma 4 E4B and E2B, and GLM-5.2 MaxMin108 for a 128 GB machine.

Custom models for your data and hardware

We fine-tune and compress models for an organisation's own data and hardware, from CPU servers to edge devices. Book a demo.

Questions, problems or a model you'd like smaller next? Reply in our feedback thread. More of our open-source work is at turintech.ai/open-source.

These are independent TurinTech releases, not affiliated with or endorsed by the original model authors. Each card links to its sources, credits and licence.