01.AI's Yi Series Large Language Models

Top Builders

Explore the top contributors showcasing the highest number of app submissions within our community.

Yi Series by 01.AI: Pushing Language Understanding Boundaries

Engineered from scratch by 01.AI developers, the Yi series models stand as a testament to cutting-edge language understanding and processing capabilities.

General
Author	01.AI
Repository	https://github.com/01-ai/Yi
Type	Large Language Models

Yi LLMs Overview

Developers can leverage two standout models within the Yi series, each offering distinct capabilities in language processing:

Yi-34B

This model flaunts exceptional performance across various evaluations such as MMLU and comprehensive LLM assessments. It outperforms larger counterparts like LLaMA2-70B and Falcon-180B while remaining cost-effective for diverse applications.

Yi-6B

With robust language processing abilities, the Yi-6B model plays a crucial role in innovative projects and diverse applications.

Opportunities

01.AI's Yi series models are engineered to cater to developers' needs, offering advanced language processing capabilities for varied applications.

01.AI, founded by AI luminary Kai-Fu Lee, introduces the Yi-34B and Yi-6B models, surpassing performance benchmarks and setting new standards in language understanding.

Reports highlight the Yi series' superior performance compared to larger pre-trained LLMs across benchmarks like common reasoning, reading comprehension, and MMLU. This exceptional performance, combined with cost-effectiveness, positions the Yi series as an optimal choice for diverse use cases.

01.AI aims to democratize AI innovation, providing open access for academic research while requiring permissions for free commercial use. As 01.AI continues to redefine language understanding, the Yi series models serve as a beacon of advancement in the AI landscape.

Yi LLMs Tutorials

Yi LLMs Resources

Edit on GitHub

Yi-LLMS AI technology page Hackathon projects

Discover innovative solutions crafted with Yi-LLMS AI technology page, developed by our community members during our engaging hackathons.

Multi-Agent-Hackathon-Suite

Hackathon Agents is a collection of three independent AI agents designed for practical, efficient AI workflows. 1. General-Purpose Agent: A cost-efficient task router that handles all required task categories using a local model by default. It automatically infers the task category from the prompt using lightweight lexical rules, avoiding model-based classification. Only when a local response fails a validity check does it escalate to the Fireworks API, minimizing paid token usage while maintaining response quality. 2. ROCm Ready: An AI-assisted CUDA-to-ROCm migration tool for AMD GPUs. It scans GitHub repositories or code snippets to detect CUDA-specific components such as kernels, runtime APIs, GPU libraries, and build configurations. Using AMD's official hipify-perl tool, it performs mechanical code conversion where possible and categorizes findings into Auto, Install-only, and Manual migration steps. A local LLM (Ollama with qwen2.5-coder:3b) generates a clear, human-readable migration report based solely on verified tool outputs. 3. Video Captioning Agent: A fully local video captioning pipeline that generates captions in four styles—formal, sarcastic, humorous_tech, and humorous_non_tech—for every input clip. The solution requires no external APIs, API keys, or cloud services, ensuring offline execution, zero usage costs, and compatibility across regions while meeting the hackathon's deployment requirements.

Token-Efficient Routing Agent

This project solves a common problem in production AI systems: using an expensive large language model for every query, even simple ones, wastes both money and compute. The Token-Efficient Routing Agent addresses this by analyzing each incoming query and automatically routing it to the most appropriate model. Simple, direct questions (e.g., "What is the capital of France?") are routed to gpt-oss-20b, a smaller and faster model. Complex questions that require reasoning, comparison, or detailed explanation (e.g., "Explain how neural networks work in detail") are routed to gpt-oss-120b, a larger and more capable model. The routing decision is based on query length and the presence of complexity-indicating keywords such as "explain," "analyze," and "compare." After each response, the agent reports the tokens used, estimated cost, and estimated savings compared to always using the large model. This approach mirrors how production AI systems can meaningfully reduce inference costs at scale without sacrificing response quality where it actually matters, making it a practical and beginner-friendly introduction to cost-aware AI system design.

Enclave: Sovereign AI That Never Leaves

Enclave is a sovereign, on-premise multi-agent AI compliance system that runs entirely inside a single AMD Instinct MI300X node, no data ever crosses a network boundary. Banks doing loan underwriting, hospitals processing patient intake, and government agencies reviewing filings need agentic AI, but sending sensitive data to OpenAI or Anthropic's cloud APIs is often a compliance violation before the first token is generated. GDPR, HIPAA, and sovereign-AI mandates require data to stay in a controlled environment. Enclave solves this with a 4-agent pipeline; Orchestrator, Researcher, Analyst, and Auditor, coordinated entirely on one GPU. This is only possible because of the MI300X's large HBM capacity: fitting multi-agent context and shared KV-cache on a single card is the difference between staying fully on-prem versus needing multiple networked GPUs, which breaks the zero-egress guarantee regulators require. We validated this on real AMD Developer Cloud hardware (ROCm 7.2, vLLM 0.16): a full compliance review pipeline ran at 32.8 tok/s with a 14.35 GiB model footprint, and a concurrent 4-agent batched execution proved 1.9x throughput versus isolated cold calls, genuine evidence of single-node concurrent execution, not just an architectural claim. Our auto-tuner enforces the single-node constraint as a hard rule: any configuration requiring a second GPU is automatically rejected, regardless of raw speed, because a network hop between GPUs is a compliance boundary crossed. Enclave turns AMD's memory advantage into a literal compliance advantage, proving that regulated industries no longer have to choose between agentic AI productivity and data sovereignty.

TokenForge — Hybrid Token-Efficient Routing Agent

The Problem Large language models are expensive and slow. Most systems send every query to an LLM — even trivial ones like "What is 12 × 4?" or "Hello!" — wasting tokens, money, and latency on tasks that never needed a model at all. There's no intelligent gatekeeper deciding which queries truly require an LLM. The Solution TokenForge (NeuroRoute) is a cost-aware routing layer that classifies each incoming query and routes it down the cheapest path that still guarantees a correct answer. Simple, deterministic queries are solved locally at zero token cost; only genuinely complex reasoning is escalated to a remote LLM. The result is the same answer quality at a fraction of the cost. How It Works A three-step pipeline: (1) Classify the query intent, (2) Route it to the optimal engine — a local AST math/logic engine or remote Fireworks AI — and (3) Execute & Measure, logging tokens saved and latency for every decision in real time Technical Implementation Local path: AST-based safe evaluation — secure parsing of math expressions with no unsafe eval(). Intent understanding: Sentence Transformers for semantic query classification. Remote path: Fireworks AI for heavy reasoning tasks. Live dashboard: An interactive interface (built on Base44) with a real-time query feed, routing decisions, latency tracking, and a cumulative token-savings counter. Impact Roughly half of all queries are resolved locally at zero token cost, with average latency in the low hundreds of milliseconds — delivering a direct, measurable reduction in LLM spend while preserving accuracy. Built for the AMD Developer Hackathon (ACT II), Track 1: Routing. Try it live: https://smart-ubiquitous-route-flow.base44.app .

DevMentor AI

🚀 DevMentor AI > **Your AI Tech Stack Advisor and Developer Mentor** DevMentor AI is a voice-enabled AI assistant that helps developers and students: * Analyze project ideas * Compare technologies * Detect redundant tools * Generate personalized learning roadmaps * Chat naturally using text or voice Built with **FastAPI**, **React**, **Gemini API**, and browser-based speech technologies. --- ## Why DevMentor AI? Students and developers often struggle to choose the right tools, avoid redundant technologies, and plan effective learning paths. DevMentor AI acts as a personalized technical mentor that helps users make better architectural decisions, understand modern technologies, and accelerate project development through natural text and voice interaction. ## ✨ Features ### 💬 AI Chat Ask technical questions and receive detailed, structured explanations. ### 🧠 Project Analyzer Describe your project idea and get: * Architecture suggestions * Recommended tech stack * Potential challenges * Improvement ideas ### ⚖️ Technology Comparison Compare tools and frameworks such as: * Electron vs Tauri * LangChain vs LangGraph * FastAPI vs Django ### 🧹 Duplicate Tool Detection Identify overlapping tools and simplify your stack. ### 🗺️ Learning Roadmap Generator Generate personalized study plans for any technical goal. ### 🎤 Voice Interaction * Speech-to-text input * Text-to-speech output * Selectable voices ### 📋 Productivity Tools * Copy responses * Download responses as Markdown * Clear current session * Persistent conversation history ### 📝 Markdown Rendering Beautiful formatting for headings, lists, tables, and code blocks.