<?xml version="1.0" encoding="UTF-8"?>
<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9" xmlns:news="http://www.google.com/schemas/sitemap-news/0.9">
<url>
<loc>https://inferencekey.com/research/memt-nano-memory-state-file</loc>
<news:news>
<news:publication>
<news:name>InferenceKey</news:name>
<news:language>en</news:language>
</news:publication>
<news:publication_date>2026-10-26</news:publication_date>
<news:title>MemT-1.0-Nano keeps a chat&apos;s memory in a 41.9 MB file, and it answered 9 of 9 questions after a reload</news:title>
</news:news>
</url>
<url>
<loc>https://inferencekey.com/research/wevisone-vl-detection-matting-one-model</loc>
<news:news>
<news:publication>
<news:name>InferenceKey</news:name>
<news:language>en</news:language>
</news:publication>
<news:publication_date>2026-10-25</news:publication_date>
<news:title>WeVisOne-VL-4B finds, counts and cuts out objects with one model, and every task ran on a real 16 GB card</news:title>
</news:news>
</url>
<url>
<loc>https://inferencekey.com/es/research/memt-nano-memory-state-file</loc>
<news:news>
<news:publication>
<news:name>InferenceKey</news:name>
<news:language>es</news:language>
</news:publication>
<news:publication_date>2026-10-26</news:publication_date>
<news:title>MemT-1.0-Nano guarda la memoria de un chat en un fichero de 41,9 MB, y acertó 9 de 9 preguntas tras recargarlo</news:title>
</news:news>
</url>
<url>
<loc>https://inferencekey.com/es/research/wevisone-vl-detection-matting-one-model</loc>
<news:news>
<news:publication>
<news:name>InferenceKey</news:name>
<news:language>es</news:language>
</news:publication>
<news:publication_date>2026-10-25</news:publication_date>
<news:title>WeVisOne-VL-4B encuentra, cuenta y recorta objetos con un solo modelo, y todas las tareas funcionaron en una tarjeta real de 16 GB</news:title>
</news:news>
</url>
<url>
<loc>https://inferencekey.com/de/research/memt-nano-memory-state-file</loc>
<news:news>
<news:publication>
<news:name>InferenceKey</news:name>
<news:language>de</news:language>
</news:publication>
<news:publication_date>2026-10-26</news:publication_date>
<news:title>MemT-1.0-Nano speichert das Gedächtnis eines Chats in einer 41,9-MB-Datei und beantwortete nach dem Neuladen 9 von 9 Fragen</news:title>
</news:news>
</url>
<url>
<loc>https://inferencekey.com/de/research/wevisone-vl-detection-matting-one-model</loc>
<news:news>
<news:publication>
<news:name>InferenceKey</news:name>
<news:language>de</news:language>
</news:publication>
<news:publication_date>2026-10-25</news:publication_date>
<news:title>WeVisOne-VL-4B findet, zählt und stellt Objekte mit einem einzigen Modell frei, und jede Aufgabe lief auf einer echten 16-GB-Karte</news:title>
</news:news>
</url>
<url>
<loc>https://inferencekey.com/fr/research/memt-nano-memory-state-file</loc>
<news:news>
<news:publication>
<news:name>InferenceKey</news:name>
<news:language>fr</news:language>
</news:publication>
<news:publication_date>2026-10-26</news:publication_date>
<news:title>MemT-1.0-Nano garde la mémoire d&apos;un chat dans un fichier de 41,9 MB, et il a répondu juste à 9 questions sur 9 après un rechargement</news:title>
</news:news>
</url>
<url>
<loc>https://inferencekey.com/fr/research/wevisone-vl-detection-matting-one-model</loc>
<news:news>
<news:publication>
<news:name>InferenceKey</news:name>
<news:language>fr</news:language>
</news:publication>
<news:publication_date>2026-10-25</news:publication_date>
<news:title>WeVisOne-VL-4B trouve, compte et détoure des objets avec un seul modèle, et chaque tâche a tourné sur une vraie carte de 16 GB</news:title>
</news:news>
</url>
<url>
<loc>https://inferencekey.com/news/memt-nano-chat-memory-reload</loc>
<news:news>
<news:publication>
<news:name>InferenceKey</news:name>
<news:language>en</news:language>
</news:publication>
<news:publication_date>2026-10-26</news:publication_date>
<news:title>MemT-1.0-Nano saves what a chat remembers to a file, and needs an 80 GB card to do it</news:title>
</news:news>
</url>
<url>
<loc>https://inferencekey.com/news/wevisone-vl-one-model-seven-jobs</loc>
<news:news>
<news:publication>
<news:name>InferenceKey</news:name>
<news:language>en</news:language>
</news:publication>
<news:publication_date>2026-10-25</news:publication_date>
<news:title>WeVisOne-VL-4B does seven vision jobs from one 16 GB card, and its fast engine miscounted a chessboard</news:title>
</news:news>
</url>
<url>
<loc>https://inferencekey.com/es/news/memt-nano-chat-memory-reload</loc>
<news:news>
<news:publication>
<news:name>InferenceKey</news:name>
<news:language>es</news:language>
</news:publication>
<news:publication_date>2026-10-26</news:publication_date>
<news:title>MemT-1.0-Nano guarda en un fichero lo que recuerda un chat, y para hacerlo necesita una tarjeta de 80 GB</news:title>
</news:news>
</url>
<url>
<loc>https://inferencekey.com/es/news/wevisone-vl-one-model-seven-jobs</loc>
<news:news>
<news:publication>
<news:name>InferenceKey</news:name>
<news:language>es</news:language>
</news:publication>
<news:publication_date>2026-10-25</news:publication_date>
<news:title>WeVisOne-VL-4B hace siete trabajos de visión en una sola tarjeta de 16 GB, y su motor rápido contó mal un tablero de ajedrez</news:title>
</news:news>
</url>
<url>
<loc>https://inferencekey.com/de/news/memt-nano-chat-memory-reload</loc>
<news:news>
<news:publication>
<news:name>InferenceKey</news:name>
<news:language>de</news:language>
</news:publication>
<news:publication_date>2026-10-26</news:publication_date>
<news:title>MemT-1.0-Nano speichert, was sich ein Chat merkt, in einer Datei, und braucht dafür eine 80-GB-Karte</news:title>
</news:news>
</url>
<url>
<loc>https://inferencekey.com/de/news/wevisone-vl-one-model-seven-jobs</loc>
<news:news>
<news:publication>
<news:name>InferenceKey</news:name>
<news:language>de</news:language>
</news:publication>
<news:publication_date>2026-10-25</news:publication_date>
<news:title>WeVisOne-VL-4B erledigt sieben Bildaufgaben auf einer 16-GB-Karte, und seine schnelle Engine hat sich bei einem Schachbrett verzählt</news:title>
</news:news>
</url>
<url>
<loc>https://inferencekey.com/fr/news/memt-nano-chat-memory-reload</loc>
<news:news>
<news:publication>
<news:name>InferenceKey</news:name>
<news:language>fr</news:language>
</news:publication>
<news:publication_date>2026-10-26</news:publication_date>
<news:title>MemT-1.0-Nano enregistre dans un fichier ce dont un chat se souvient, et il lui faut une carte de 80 GB pour ça</news:title>
</news:news>
</url>
<url>
<loc>https://inferencekey.com/fr/news/wevisone-vl-one-model-seven-jobs</loc>
<news:news>
<news:publication>
<news:name>InferenceKey</news:name>
<news:language>fr</news:language>
</news:publication>
<news:publication_date>2026-10-25</news:publication_date>
<news:title>WeVisOne-VL-4B fait sept métiers de vision sur une seule carte de 16 GB, et son moteur rapide a mal compté un échiquier</news:title>
</news:news>
</url>
</urlset>
