Keep model families and distribution links current on the organization page

README.md

@@ -9,15 +9,19 @@ pinned: false
 
 # TurinTech models
 
-TurinTech models, built with Artemis. Smaller GGUF versions of open models, each tested against the model it came from and run on stock llama.cpp.
+TurinTech models, built with Artemis. GGUF versions of open models, with downloadable files, measured comparisons and runtime instructions on each model card.
 
 ## Download decision models
 
-[Compressed decision models](https://huggingface.co/collections/turintech/compressed-decision-models-6accea74eb7016a886f25fdb): Clef, Clef-Flash, Perplexity Decider 27B, Kev-4B, Kev-9B, Kev-27B, d3-flash and d3 27B. 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.
+[Compressed decision models](https://huggingface.co/collections/turintech/compressed-decision-models-6accea74eb7016a886f25fdb): Clef, Kev, Perplexity Decider and d3. For structured decisions and answer probabilities, follow each card's tested `/v1/systemone` endpoint and settings.
 
 ## Download compressed chat models
 
-[Compressed LLMs](https://huggingface.co/collections/turintech/compressed-llms-6acd251457ba2b9f33ad9d8a): Qwen3.5 0.8B, Gemma 4 E4B and E2B, and GLM-5.2 MaxMin108 for a 128 GB machine.
+[Compressed LLMs](https://huggingface.co/collections/turintech/compressed-llms-6acd251457ba2b9f33ad9d8a): Qwen3.5, Gemma 4 and GLM-5.2. Browse the collection for published sizes and variants, with measured trade-offs on each card.
+
+## Ollama
+
+Available native packages are listed on [Ollama](https://ollama.com/turintech). Follow the model card for supported runtimes, recommended tags and run commands.
 
 ## Artemis
 

index.html

@@ -9,11 +9,13 @@
 	<body>
 		<div class="card">
 			<h1 style="font-size:18px;font-weight:700;margin:0 0 6px 0"><strong>TurinTech models</strong></h1>
-			<p style="margin:0 0 8px 0">TurinTech models, built with Artemis. Smaller GGUF versions of open models, each tested against the model it came from and run on stock llama.cpp.</p>
+			<p style="margin:0 0 8px 0">TurinTech models, built with Artemis. GGUF versions of open models, with downloadable files, measured comparisons and runtime instructions on each model card.</p>
 			<h2 style="font-size:16px;font-weight:600;margin:16px 0 4px 0"><strong>Download decision models</strong></h2>
-			<p style="margin:0 0 8px 0"><a href="https://huggingface.co/collections/turintech/compressed-decision-models-6accea74eb7016a886f25fdb" target="_blank" style="color:#2563eb;text-decoration:underline">Compressed decision models</a>: Clef, Clef-Flash, Perplexity Decider 27B, Kev-4B, Kev-9B, Kev-27B, d3-flash and d3 27B. They return a probability for each allowed answer through llama.cpp's <code>/v1/systemone</code> endpoint and are not chat models. Start with Clef-Flash (4.9 GB), the smaller Clef.</p>
+			<p style="margin:0 0 8px 0"><a href="https://huggingface.co/collections/turintech/compressed-decision-models-6accea74eb7016a886f25fdb" target="_blank" style="color:#2563eb;text-decoration:underline">Compressed decision models</a>: Clef, Kev, Perplexity Decider and d3. For structured decisions and answer probabilities, follow each card's tested <code>/v1/systemone</code> endpoint and settings.</p>
 			<h2 style="font-size:16px;font-weight:600;margin:16px 0 4px 0"><strong>Download compressed chat models</strong></h2>
-			<p style="margin:0 0 8px 0"><a href="https://huggingface.co/collections/turintech/compressed-llms-6acd251457ba2b9f33ad9d8a" target="_blank" style="color:#2563eb;text-decoration:underline">Compressed LLMs</a>: Qwen3.5 0.8B, Gemma 4 E4B and E2B, and GLM-5.2 MaxMin108 for a 128 GB machine.</p>
+			<p style="margin:0 0 8px 0"><a href="https://huggingface.co/collections/turintech/compressed-llms-6acd251457ba2b9f33ad9d8a" target="_blank" style="color:#2563eb;text-decoration:underline">Compressed LLMs</a>: Qwen3.5, Gemma 4 and GLM-5.2. Browse the collection for published sizes and variants, with measured trade-offs on each card.</p>
+			<h2 style="font-size:16px;font-weight:600;margin:16px 0 4px 0"><strong>Ollama</strong></h2>
+			<p style="margin:0 0 8px 0">Available native packages are listed on <a href="https://ollama.com/turintech" target="_blank" style="color:#2563eb;text-decoration:underline">Ollama</a>. Follow the model card for supported runtimes, recommended tags and run commands.</p>
 			<h2 style="font-size:16px;font-weight:600;margin:16px 0 4px 0"><strong>Artemis</strong></h2>
 			<p style="margin:0 0 8px 0">Built with Artemis, TurinTech's optimisation platform. Teams can use Artemis to make their own models smaller and faster, or to optimise their code. More at <a href="https://www.turintech.ai" target="_blank" style="color:#2563eb;text-decoration:underline">turintech.ai</a>, or <a href="https://www.turintech.ai/book-a-demo" target="_blank" style="color:#2563eb;text-decoration:underline">book a demo</a>.</p>
 			<p style="margin:0 0 8px 0">Questions, problems or a request for the next model to shrink? Reply in our <a href="https://huggingface.co/spaces/turintech/README/discussions/1" target="_blank" style="color:#2563eb;text-decoration:underline">feedback thread</a>. More of our open-source work is at <a href="https://www.turintech.ai/open-source" target="_blank" style="color:#2563eb;text-decoration:underline">turintech.ai/open-source</a>.</p>