https://ai.ngchi.online/tags/ai-models/ Recent content in AI Models on AI Notes Hugo en-us Mon, 28 Sep 2026 15:17:11 +0700 https://ai.ngchi.online/posts/laya-vs-jev-a-challenger-with-its-own-scorecard/ Mon, 28 Sep 2026 15:17:11 +0700 https://ai.ngchi.online/posts/laya-vs-jev-a-challenger-with-its-own-scorecard/ An open-source challenger to Jev ships with a comparison table. The accuracy claim does not survive reading the rest of the page, but the structural argument does. https://ai.ngchi.online/posts/gpt-6-sol-and-luna-reading-the-benchmark-footnotes/ Thu, 24 Sep 2026 17:21:29 +0700 https://ai.ngchi.online/posts/gpt-6-sol-and-luna-reading-the-benchmark-footnotes/ OpenAI’s GPT-6 Sol and Luna announcement is a cost-efficiency pitch. Its own footnotes concede that the closest competitor’s cost is the one figure not reported. https://ai.ngchi.online/posts/what-people-are-actually-building-on-jev/ Wed, 23 Sep 2026 14:35:59 +0700 https://ai.ngchi.online/posts/what-people-are-actually-building-on-jev/ A week after launch is when the real question gets answered: does anyone build on it. For Jev the early answer is code review, model routing and linting - places where which one matters more than write me something. https://ai.ngchi.online/posts/jev-and-typesafe-ais-system-one-models/ Wed, 23 Sep 2026 14:35:47 +0700 https://ai.ngchi.online/posts/jev-and-typesafe-ais-system-one-models/ Every model you have used generates text one token at a time, even when all you needed was yes or no. TypeSafe AI built one that returns typed values instead. https://ai.ngchi.online/posts/ternary-bonsai-2-27b-a-27b-model-in-6gb/ Wed, 23 Sep 2026 14:35:36 +0700 https://ai.ngchi.online/posts/ternary-bonsai-2-27b-a-27b-model-in-6gb/ A 27B model squeezed into 5.95GB at 1.72 bits per weight, claiming 98.2% of full-precision performance on its own benchmark suite - and what that compression actually costs. https://ai.ngchi.online/posts/qwen-image-21-native-transparency-and-a-license-catch/ Wed, 23 Sep 2026 14:27:26 +0700 https://ai.ngchi.online/posts/qwen-image-21-native-transparency-and-a-license-catch/ Alibaba built transparency into the image format itself rather than removing backgrounds afterwards - then shipped it under a research licence instead of the Apache 2.0 its predecessors used. https://ai.ngchi.online/posts/deepseek-v41-flash-and-the-kv-cache-problem/ Wed, 23 Sep 2026 14:27:17 +0700 https://ai.ngchi.online/posts/deepseek-v41-flash-and-the-kv-cache-problem/ DeepSeek-V4.1-Flash is bigger than its predecessor and cheaper to run. The savings come from cutting KV cache to 890 bytes per token - about a quarter of the previous model.