AI directory
Only well-known products. Each page has an intro, publicly stated direction, and a dated version timeline. Dates come from public reports; official docs win if they disagree.
Compiled through 2026-09
Pick by task6 scenesOptional: primary picks by task · Trae · Claude · Jimeng · Kling
Optional: primary picks by task · Trae · Claude · Jimeng · Kling
Start from what you need to get done. Each row lists one primary pick plus alternatives.
Write & refactor code
Multi-file edits and repo-scale refactors — pick an agentic editor
TraeWhy this pick：Works in China with a generous free tier and the most complete agentic toolkit among domestic editors
Docs, reports & long writing
Long-form work rewards instruction-following and a big context window
ClaudeWhy this pick：Consistently top-tier instruction following for long-form writing; overseas access needed — Kimi / DeepSeek are solid domestic fallbacks
Images, posters & ad creative
For Chinese prompts and local commerce styles, domestic tools save time
JimengWhy this pick：Best Chinese-prompt understanding with built-in e-commerce templates; direct access and free tier in China
Video & short-form clips
Camera control and character consistency matter most here
KlingWhy this pick：First-tier camera control and character consistency, direct China access, cheaper tiers than Runway
Research with citations
Insist on clickable sources rather than trusting a bare answer
PerplexityWhy this pick：The most complete citation and follow-up experience; overseas access needed — Felo / Kimi work directly in China
Local & self-hosted
Data stays in-house with no metered cost — but you run the hardware
DeepSeekWhy this pick：Open weights with the lowest VRAM bar in its class; or skip self-hosting via the official app/API
Editor's shortlists
Short lists filtered by different dimensions. Click through for details.
Top rated
By editorial score
Shipping fastest
Most releases in the past year
Works in China
Start using without an overseas network
Free & open source
Free tier or open weights
Chat models
19Consumer and enterprise chat: writing, reasoning, and multimodal work.
ChatGPT
OpenAI·2015
The world's leading AI platform, upgraded to GPT-6 with adaptive reasoning and autonomous multi-agent execution.
Claude
Anthropic·2021
Known for writing quality, long context, and coding agents. Claude Code, Artifacts, and computer use are the differentiators.
Gemini
Google·2015
Google’s front door across Search, Gmail, Docs, YouTube, and huge context. The Flash line iterates very fast.
Grok
xAI·2023
xAI’s flagship: realtime info, a sharp tone, and coding. Now in Cursor, GitHub Copilot, and Grok Build.
DeepSeek
DeepSeek·2023
China’s flagship open-weight + API lab. R1 put it on the global map; still extreme value.
Qwen
Alibaba / Alibaba Cloud·2023
Alibaba’s prolific open-weight family: chat, code, multimodal, and DashScope / Model Studio.
GLM
Zhipu AI·2019
Strong on China enterprise and coding. Qingyan app plus an API, aimed at long-horizon work.
Kimi
Moonshot AI·2023
Started with huge context; K2/K3 moved into open flagship models and long coding. Kimi Code and Kimi Work complete the suite.
Doubao
ByteDance·2023
ByteDance’s consumer front door—huge traffic in China. Chat, office, image, and video (Jimeng, Seedance).
MiniMax
MiniMax·2021
China frontier lab famed for speech synthesis, long context, and Hailuo AI consumer platform.
StepFun
StepFun·2023
Hardcore AI lab behind Step-1/Step-2 trillion-parameter MoE models and Yuewen productivity assistant.
ERNIE
Baidu·2019
China's early conversational LLM pioneer, integrated with Baidu Search and Qianfan platform.
Llama
Meta·2023
Meta's open-weight model family with the world's largest self-hosting and fine-tuning ecosystem.
Mistral
Mistral AI·2023
France's AI star: open-weight models with best-in-class small-model efficiency and EU compliance.
Microsoft Copilot
Microsoft / OpenAI·2023
Microsoft's fleet-wide assistant bringing GPT into Windows, Office and the browser.
Amazon Nova
AWS·2024
AWS's in-house multimodal family, delivered via Bedrock with enterprise cost efficiency.
Hunyuan
Tencent·2023
Tencent's in-house LLM, leveraging the WeChat ecosystem and open video/3D model lines.
Hailuo AI
MiniMax·2024
MiniMax's consumer assistant, standing out in voice companionship and video generation.
iFlytek Spark
iFlytek·2023
iFlytek's cognitive LLM, anchored in speech interaction plus education and government verticals.
Coding agents
11Coding agents in the editor and terminal for daily and repo-scale work.
OpenAI Codex
OpenAI·2021
OpenAI's foundational coding model line, driving GitHub Copilot and ChatGPT Canvas.
Cursor
Anysphere·2022
A default AI code editor. Tab, Composer, Agent, MCP, and model picking are the core.
Windsurf
Codeium·2021
Next-gen AI IDE by Codeium featuring Cascade flows that keep human and agent in lockstep.
Trae
ByteDance·2024
ByteDance's native AI IDE featuring powerful Builder agent mode and generous flagship model access.
v0
Vercel·2023
Vercel's generative UI platform that turns natural language into clean React and Tailwind code.
GitHub Copilot
GitHub / Microsoft·2021
The widest coding assistant: VS Code, JetBrains, CLI, cloud agents, mature enterprise buy.
Claude Code
Anthropic·2025
A terminal coding agent: read the repo, edit, test. Tightly bound to Claude models and MCP.
Devin
Cognition·2024
The first autonomous AI software engineer. Operates in a sandbox with shell and browser to solve full SWE tasks.
Cline
Open Source / Cline·2024
The premier open-source autonomous coding agent for VS Code, powered by deep MCP toolchain support.
Bolt.new
StackBlitz·2024
In-browser AI web dev platform. Build, run, edit, and deploy full-stack applications with natural language.
Manus
Monica Team·2025
Next-gen general autonomous agent. Excels at complex market research, data analysis, and multi-step web tasks.
Image & video
7Text-to-image, image-to-video, and widely used creator tools.
Midjourney
Midjourney, Inc.·2022
A taste benchmark for image gen. Started on Discord, then Web; default around V8.2 in 2026.
FLUX.1
Black Forest Labs·2024
Frontier image foundation models by former Stable Diffusion creators, excelling at prompt adherence & text.
Kling
Kuaishou·2024
A leading China video model. Duration, camera motion, and character consistency have a strong reputation.
Jimeng
ByteDance·2024
CapCut-family AI creation. Images, posters, and video tightly tied to China creator workflows.
Suno
Suno, Inc.·2023
Global leader in generative music, turning simple text into broadcast-ready full songs.
Runway
Runway·2018
The industry standard for cinematic AI video generation, renowned for camera control and visual effects tooling.
Luma AI
Luma·2021
High-fidelity AI video generator known for realistic world physics, smooth camera sweeps, and keyframing.
Search & research
5Cited answer engines for research, comparison, and tracking public info.
Perplexity
Perplexity AI·2022
Cited AI search. Built for research, comparison, and news—not closed-door writing.
NotebookLM
Google·2023
Google's personalized AI research assistant grounded in your notes, with viral Audio Overviews.
Genspark
MainFunc·2024
Multi-agent autonomous search engine that generates custom, structured, ad-free research pages on the fly.
Felo AI
Sparticle·2024
Cross-lingual AI search powerhouse that breaks language barriers to summarize global sources and generate mind maps.
Nami AI
360 Group·2024
360's multi-model AI search, free cross-model answers and one-click deep research reports.
Release feed
Recently shipped versions across all listed products, newest first.
Next-gen multimodal reasoning & autonomous agent flagship.
1M context by default; adaptive thinking.
Shipped Sep 2; stronger docs/video/retrieval.
Flagship vs high-QPS split in early Sep.
1M context default for huge repos.
Rapid Flash cadence; lower cost.
Aug 12 launch; Copilot on Aug 14.
August release; coding + multimodal Flash.
Frequently asked
›Which AI tools work in mainland China?
16 products are marked "Works in China": DeepSeek, Qwen, GLM, Kimi, Doubao, MiniMax, StepFun, ERNIE, and more. They need no extra network setup to sign up and use.
›Which AI tools have a free tier?
37 products offer a free tier or are open source, including ChatGPT, Claude, Gemini, Grok, DeepSeek, Qwen, GLM, Kimi, and more. Free tiers usually come with rate limits — check official pricing before relying on them.
›Which models can be self-hosted or deployed on-prem?
8 products ship open weights and can run locally: DeepSeek, Qwen, GLM, Kimi, FLUX.1, Cline, Llama, Hunyuan. Self-hosting removes metered cost and keeps data in-house, but you supply the hardware and ops.
›How are the ratings calculated?
Ratings are an editorial reference combining public benchmarks, community feedback, and hands-on impressions. They are not a third-party benchmark verdict. Test against your own workload before buying.
Timelines are compiled from public sources and may lag. Check official sites, pricing, and changelogs before you buy or integrate. Not investment advice.