https://mypocketai.app/blog On-device AI, open models and iPhone hardware. en-us Sat, 12 Sep 2026 10:19:43 GMT https://mypocketai.app/blog/spark-x2-5-4b-iphone https://mypocketai.app/blog/spark-x2-5-4b-iphone Sat, 12 Sep 2026 09:00:00 GMT iFLYTEK's Ciyuan Xinghuo open-sourced a 1.7B and a 4B dense model on September 1 under Apache 2.0 — the 1.7B fits down to a 4 GB iPhone, the 4B needs 8 GB, and the advertised 1M-token context is not something either one can actually use on a phone. https://mypocketai.app/blog/best-offline-ai-chat-apps-for-iphone https://mypocketai.app/blog/best-offline-ai-chat-apps-for-iphone Thu, 10 Sep 2026 09:00:00 GMT Six iPhone apps that run a real LLM on-device with no internet — Enclave, Locally AI, Private LLM, PocketPal, Apollo and Pocket AI — compared on price, app size, iOS requirement and what they actually do offline. https://mypocketai.app/blog/minicpm5-2b-iphone https://mypocketai.app/blog/minicpm5-2b-iphone Wed, 09 Sep 2026 09:00:00 GMT OpenBMB's MiniCPM5-2B shipped September 7 as a plain Llama-architecture checkpoint — no fork, no waiting on llama.cpp support — and it is small enough to sit comfortably on a 6 GB iPhone. https://mypocketai.app/blog/k2-horizon-mbzuai-iphone https://mypocketai.app/blog/k2-horizon-mbzuai-iphone Tue, 08 Sep 2026 09:00:00 GMT MBZUAI's IFM released a 3.7B and a 7B model built for phones on September 3, but upstream llama.cpp does not support the architecture yet, so nothing built on stock llama.cpp — Pocket AI included — can load them today. https://mypocketai.app/blog/how-much-ram-does-an-llm-need-on-iphone https://mypocketai.app/blog/how-much-ram-does-an-llm-need-on-iphone Sun, 06 Sep 2026 09:00:00 GMT A 2 GB model does not need 2 GB of RAM. Here is the arithmetic we actually enforce in Pocket AI, the incident that produced it, and the number your iPhone can really hold. https://mypocketai.app/blog/which-ai-models-run-on-your-iphone https://mypocketai.app/blog/which-ai-models-run-on-your-iphone Sat, 05 Sep 2026 09:00:00 GMT A model-by-model reference for on-device LLMs — download size, the RAM each one really needs, and context length. 34 open models, sorted by the phone that can hold them. https://mypocketai.app/blog/gguf-quantization-q4-k-m-explained https://mypocketai.app/blog/gguf-quantization-q4-k-m-explained Fri, 04 Sep 2026 09:00:00 GMT What the letters in Q4_K_M actually mean, why almost every phone model uses it, and the one case where we deliberately ship Q8_0 instead.