A family of open code models (1.3B-33B) trained from scratch on 2T tokens of project-level code, using a 16K-window fill-in-the-blank objective. Beats Codex and GPT-3.5 on code benchmarks and ships under a license permitting commercial use.
AI Agent2024
A family of GUI agents that operate phones, desktops, and browsers by perceiving the screen visually rather than reading app code. Ships open GUI-Owl vision-language models (7B/32B) plus a multi-agent framework for planning, reflection, and tool use.
AI Model2024
Pocket-sized multimodal LLM for efficient image- and video-understanding on mobile and edge devices, featuring mixed 4x/16x visual-token compression (MiniCPM‑V 4.6), compact 1.3B variants, and ready guides for iOS/Android/HarmonyOS deployment.
Chatbot2024
Creates personalized digital avatars (AI twins) by fine-tuning LLMs on users' chat history and binding them to chatbots. Provides an end-to-end pipeline — chat export, preprocessing with privacy filters, SFT/LoRA training, and deployment (Telegram/Discord/Slack). Best with larger models and substantial chat data.
AI Train2024
Runs GPT-4o-class vision, speech, and full-duplex audio-video conversation on a 9B model small enough to deploy on phones and tablets. The 4.5 release scores 77.6 on OpenCompass and adds real-time bilingual voice with voice cloning.
AI Deploy2024
GPU kernel library for LLM inference attention, sampling, and KV-cache, built on block-sparse formats with JIT-compiled customizable templates. Reports 29-69% inter-token-latency cuts vs compiler backends; powers SGLang, vLLM, and MLC-Engine.
Reaches 51.7% on the competition-level MATH benchmark with a 7B model and no tools or voting, rivaling Gemini-Ultra and GPT-4. Built on a 120B-token math corpus mined from Common Crawl, and introduces GRPO, a memory-efficient PPO variant for reasoning.
AI Client2024
Self-hostable “bookmark everything” app for saving links, notes, images and PDFs with automatic fetching of previews, full-text search, OCR, and LLM-based automatic tagging and summarization (supports local models via ollama). Targets users who want AI-assisted organization in a self-hosted stack.
AI Infra2024
Builds a knowledge graph from a text corpus by extracting entities and relations, clusters it into communities with the Leiden algorithm, and summarizes them — so queries can synthesize across scattered documents instead of retrieving isolated chunks.
AI Agent2024
Controls customer-facing LLM agents turn-by-turn against deterministic guidelines instead of one big system prompt, surfacing only the rules and tools that apply each turn. Adds journeys, pre-approved canned responses, and traces for auditable behavior.
