AI tools¶

Language models¶

OpenAI
- GPT 4o: General purpose use, can search the web, can add context to the conversation.
- GPT 5: Latest model with thinking mode.
Claude: excels in coding, has a smaller ecosystem with fewer third-party integrations and plugins.
Gemini: Longer context, thinking mode.
LLaMA: Open source LLM developed by Meta.
DeepSeek: The recent DeepSeek R1 + Search is comparable to GPT o1.
Grok: xAI's LLM with thinking mode as well, can search the internet.

The table below may be out-of-date. Please check the latest models from the official websites.

Need	Recommended Models	Key Features
High-Quality, Nuanced Output	GPT 5, Gemini 2.5 Pro, DeepSeek R1, Claude 4 Sonnet, Grok 4	Top-tier quality, well-structured responses, ideal for professional applications
Speed-Sensitive Applications	Gemini 2.5 Flash	Fastest output speeds, minimal delay, great for real-time tasks
Cost-Conscious Deployments	Mistral 3B, Llama 4	Highly affordable per million tokens, cost-effective for large-scale use
Extensive Context Processing	Gemini 2.5 Pro	2 million token context window, supports deep contextual analysis
Low-Latency, Fast Response	Mistral NeMo	Minimal response time, excellent for interactive, real-time applications

Explore and choose the best language model for your tasks, e.g., creative writing, coding, math, etc.

If you are Harvard affiliated, you can get access to many language models through Harvard AI Sandbox.

Check Chat Arena for the assessment of the language models under different tasks.

Assistant Code¶

GitHub Copilot: IDE based, copilot based, offers autocompletion, code generation, and code review.
Cursor: IDE based, autocompletion, code generation, and agentic features for automating complex workflows.
Google Antigravity: Google’s agentic development platform for agents to plan, execute, and verify complex tasks.
Codex: OpenAI’s coding agent/API for generating and editing code, useful for building automated coding workflows.
Windsurf: IDE based, positions itself as an "agentic" IDE, cascade mode for generating entire processes based on prompts.
Tabnine: IDE based
Codeium: Developed by the same team as Windsurf, offers autocompletion, code generation, and code review. It can be extensions to your current IDE like VS Code.
Devin AI: Web based, Operates autonomously, capable of executing entire development tasks without human intervention, effectively acting as an independent software engineer rather than a coding assistant. Operates within a secure, sandboxed environment equipped with common developer tools. $500/month.
Claude: Web based, provides implementation of code after code generation.
Cline: Open sourced tool to code as IDE extension.

Agentic Coding¶

There are more agentic coding tools which can run commands in terminal. You can design a todo list and it will generate pipelines of commands using tools to achieve your goals. There are two types of agentic coding tools now: IDE-based and terminal-based. All the IDE-based tools we listed above have the agentic features as well, especially for Cursor, Codex and Antigravity. Below we list some terminal-based agentic coding tools.

Claude Code: One of the best terminal coding tools.
Gemini CLI: Similar to Claude Code, but with free tier: 60 requests/min and 1,000 requests/day with personal Google account.
Qwen Code: Based on the Qwen model.

Agentic AI¶

OpenClaw: OpenClaw is a platform for building and running AI agents to control your entire system instead of a single project.

Data analysis¶

Pardus AI: Prompt-free data analysis tool. Just upload your data and it will generate the analysis report automatically.

Search tools¶

Perplexity: delivering real-time, well-referenced information.
ChatGPT Search: Search within the ChatGPT interface.
Deep research: many language models can the deep research tools and they are great for literature review or idea brainstorming. For GPT models, there is Deep Research. For Gemini, there is Deep Research. For Grok, there is DeepSearch.

Notebook and Knowledge Management¶

Notion AI: An all-in-one workspace that combines note-taking, project management, and databases. Its integrated AI assistant helps with summarizing texts, generating content ideas, and improving writing quality.
Google NotebookLLM: AI-powered research and note-taking tool that assists users in interacting with their documents. It can generate summaries, explanations, and answers based on uploaded content, enhancing understanding and organization.
ClickUp: Primarily a project management tool, ClickUp incorporates AI to enhance note-taking and task management. It offers features like AI-generated summaries and task suggestions, streamlining workflows
OneNote with Copilot: OneNote integrates a range of AI features through Microsoft’s own 365 Copilot, enhancing the way you create and manage your notes. That includes document generation, text summarization, outlining, and even an AI chat.

Writing tools¶

Grammarly: A widely used AI-powered writing assistant that provides real-time grammar and spelling corrections, style suggestions, and clarity improvements.
Quillbot: It offers a suite of tools tailored to assist users in various aspects of writing, including paraphrasing, grammar checking, summarizing, and more.
GPT Zero: Detect AI generated text.
Gamma: AI generated slides.

Transcription and meeting assistants¶

Zoom AI Companion: can take notes, summarize meetings, create action items, and analyze the predominant tones and speakers during calls
Fireflies: helps transcribe, summarize, record, filter, and analyze virtual meetings and conversations. It integrates with platforms like Zoom, Microsoft Teams, Google Meet, and Webex.
tl;dv: distinguishes itself with multi-meeting AI insights and extensive integrations

Image/video generation¶

Nano Banana: Google's image generation tool. You can directly use it in Gemini.
Midjourney: Renowned for producing high-quality, artistic images, Midjourney is popular among creators for its unique aesthetic outputs. It operates via a Discord-based interface, facilitating community engagement and prompt sharing.

Stable Diffusion: An open-source model by Stability AI, Stable Diffusion allows users to generate images on local machines, offering flexibility and control. Its open-source nature encourages community-driven enhancements and customizations
Jimeng: Powered by Seedance series models. Need Chinese phone number or tiktok account to register.

Voice and music generation¶

NotebookLM: Can generate podcast, slides, and more based on your uploaded documents.
Eleven Labs: AI-assisted text-to-speech software capable of producing lifelike speech by synthesizing vocal emotion and intonation. It also provides multilingual speech generation and long-form content creation with contextually-aware voices.
Wondercraft: generates scripts and offers a vast library of hyper-realistic AI voices to host your show.
Descript: An all-in-one editing tool for both audio and video, Descript uses AI to automate tasks like noise reduction and audio leveling. It also offers features like overdubbing, which allows for voice cloning, and the ability to remove filler words from your audio.
Podcastle: An all-in-one editing tool for both audio and video, Descript uses AI to automate tasks like noise reduction and audio leveling. It also offers features like overdubbing, which allows for voice cloning, and the ability to remove filler words from your audio
Suno: Powerful AI song and music generator.

Prompt Engineering¶

PromptPerfect: transforms user ideas into precise prompts suitable for various AI models, including GPT-4, Midjourney, and Claude. It supports both text and image models
prompts.chat: Prompt examples for various tasks.
Copy Coder: Upload images of full applications, UI mockups, or custom designs and it will generate prompts to build your apps faster.