Research & Local AI

AutoGen

Microsoft open-source multi-agent framework (AgentChat/Core/Studio). Free OSS; maintenance mode—new builds: Microsoft Agent Framework. LLM costs separate.

ChatGPT

OpenAI’s AI assistant for chat, research, coding (Codex), images, and agents—Free, Go ($8), Plus ($20), Pro ($100/$200), Business, and Enterprise plans.

Claude

Anthropic's AI assistant for chat, coding, research, and agents. Free, Pro ($20), Max ($100–$200), Team, Enterprise, plus Claude API.

Claude 3 Opus

Claude Managed Agents

Anthropic’s hosted Claude agent runtime (public beta Apr 2026): sandboxes, sessions, tools/MCP, tokens + $0.08/session-hour—not Claude Code or the Agent SDK.

Claude Projects

Anthropic’s Claude.ai workspaces with project knowledge, custom instructions, RAG-scale docs, and Team sharing—organize chats without re-pasting context.

CrewAI

Open-source multi-agent Python framework (MIT, ~55.7k GitHub stars) plus AMP platform: Crews, Flows, Studio. Free OSS/cloud (50 executions); Enterprise custom.

Databutton

Databutton (now Riff): AI full-stack app builder—Python FastAPI + React, plan-then-build agent, built-in auth/DB, one-click deploy. Credits from ~$20/mo; rebranded enterprise Riff.

DeepSeek

DeepSeek V4: free chat, OpenAI-compatible API, MIT open weights. V4-Flash & V4-Pro with 1M context, thinking modes, and aggressive cache pricing.

Dify

Open-source LLM app platform (LangGenius): visual workflows, RAG, agents. Cloud Sandbox free; Pro $59/mo; Team $159/mo; Community self-host free.

Emergent AI

YC S24 AI app builder (emergent.sh): multi-agent chat to full-stack web & mobile apps with deploy, GitHub, credit plans Free–Pro $20–$200/mo.

Exa

Neural web search API for AI agents: Search $7/1k, Deep $12–15, Contents $1/1k, Agent from $0.012/run—plus MCP, Websets, Monitors.

Flowise

Open-source Apache 2.0 visual builder for LLM agents and RAG (YC S23; Workday-acquired). Self-host free; Cloud Free–Pro from $0–$65/mo.

Gemini

Google’s multimodal AI assistant and model family: free Gemini app, Google AI Plus/Pro/Ultra plans, and pay-as-you-go Developer API for chat, research, coding, and media.

Gemini Code Assist

Google’s seat-licensed AI coding assistant (Standard $19–$22.80/user, Enterprise $45–$54/user). Consumer free/Pro/Ultra IDE & CLI paths ended June 18 2026 for Antigravity.

Google Colab

Google’s hosted Jupyter notebooks with free GPUs/TPUs. Pro $9.99/mo (100 compute units); Pro+ $49.99/mo (600 units); Pay As You Go packs. Best for learning and short ML jobs—not guaranteed long training.

Google Gemma 4

Google DeepMind open-weight multimodal LLMs (E2B–31B) under Apache 2.0: reasoning, agents, 256K context, free weights, local + Vertex AI.

Grok

xAI Grok: SuperGrok chat, Grok 4.5 API, Imagine, Voice, and Grok Build—frontier models with real-time web/X tools and freemium pricing.

Jan

Open-source ChatGPT alternative (Apache 2.0) by Menlo Research. Local LLMs + cloud APIs, MCP, CLI, OpenAI-compatible server on localhost:1337.

Kimi

Moonshot AI’s Kimi stack: freemium chat/agents, Kimi Work, Kimi Code CLI, and OpenAI-compatible API. Flagship K3 (2.8T MoE, 1M context) plus K2.x coding models.

Kimi K3

Moonshot AI’s July 2026 flagship: 2.8T MoE, 1M context, native multimodal, $3/$15 API, open weights targeted ~27 Jul 2026.

LangChain

Open-source MIT framework for LLM agents (create_agent, LangGraph) plus LangSmith: free Developer tier, Plus $39/seat, Enterprise custom.

LangFuse

Open-source LLM engineering platform for tracing, prompts, evals, and cost monitoring. Free MIT self-host; Cloud Hobby free; Core $29/mo.

LangGraph

MIT-licensed low-level agent orchestration by LangChain Inc: stateful graphs, checkpoints, HITL, streaming. Free OSS; LangSmith Plus from $39/seat for managed deploy.

LlamaIndex

Open-source data framework for LLM apps over your docs, plus LlamaParse credits for agentic OCR, extraction, and managed indexes. Free OSS; Starter $50.

LM Studio

Free desktop app (Element Labs) to discover, download, and run local LLMs (GGUF/MLX). OpenAI-compatible API, MCP, CLI/llmster. Free for personal and work use.

Manus

General AI agent with a cloud computer for research, browser automation, slides, and web apps. Freemium credits; Pro from $20/mo.

Meta Llama 3

Meta’s open-weight Llama 3.x and Llama 4 (Scout/Maverick) models: free downloads under Community License, multimodal MoE, and hosted APIs via Groq, Bedrock, Azure, and others.

Mistral

Mistral AI: open-weight and commercial LLMs, Vibe agent (ex-Le Chat), and La Plateforme API. Free, Pro $14.99, Team $24.99/user, Enterprise; EU-hosted.

NotebookLLM

Google's source-grounded AI research notebook (also Gemini Notebook): upload PDFs/Drive/YouTube, chat with citations, Audio/Video Overviews, free + Google AI plans.

NotebookLM

Google’s source-grounded AI research notebook (also marketed as Gemini Notebook). Free tier + Google AI Plus/Pro/Ultra; Enterprise ~$9/license. Audio/Video Overviews, citations.

Ollama

Free local LLM runtime (MIT) with optional Ollama Cloud (Pro $20, Max $100). OpenAI- & Anthropic-compatible API, model registry, agents.

Open WebUI

Self-hosted ChatGPT-style AI interface for Ollama, OpenAI, Anthropic & compatible APIs. Free with branding; enterprise for white-label. RAG, tools, multi-user RBAC.

OpenAI Operator

OpenAI’s retired browser agent (Jan 2025); folded into ChatGPT agent, then Work/cloud browser. CUA-powered web tasks via paid ChatGPT plans.

Paperclip

Open-source MIT control plane for AI agent teams: org charts, budgets, heartbeats, tickets, board governance. Self-host free; pay infra + LLM usage.

Perplexity

AI answer engine with live web citations, Deep Research, multi-model Pro/Max, Computer agents, Comet browser, and Sonar/Search/Agent APIs.

Perplexity AI API

Perplexity’s developer platform: Agent API (multi-provider models + web tools), Search API, Sonar chat, and Embeddings—pay-as-you-go with citations and SOC 2.

Qwen

Alibaba's Qwen LLM family: free Qwen Studio chat, Apache open weights (Qwen3.6), and Model Studio APIs with Turbo/Plus/Max pricing and Coding Plan.

Qwen 3.6 Dense

Alibaba Qwen open-weight dense 27B multimodal LLM (Apr 2026): Apache 2.0, hybrid Gated DeltaNet, 262K–1M context, flagship agentic coding for local & API use.

Runway

Runway is a creative AI platform for video/image generation (Gen-4.5, Aleph 2.0), multi-model workspace, API, and Characters. Credit-based plans from free to Max/Enterprise.

Runway Gen-5

Runway’s flagship AI video model is Gen-4.5 (Dec 2025)—not Gen-5. Credit-based plans from $12/mo, API at $0.12/sec, director controls, Aleph/Act-Two on the same platform.

Sol

OpenAI’s GPT-5.6 flagship model (Sol): ChatGPT reasoning + API at $5/$30 per 1M tokens. Coding agents, max/ultra effort, Plus $20 entry—not Solana.

Veo 3

Google DeepMind’s 2025 native-audio AI video model (Veo 3 / Veo 3 Fast). ~8s clips via Gemini, Flow, Gemini API & Vertex; 3.0 API IDs shut down June 30, 2026—migrate to Veo 3.1.

Veo 3.1

Google DeepMind’s flagship AI video model: 8-second clips with native audio, 720p–4K, image/reference controls, available in Gemini, Flow, Gemini API, and Vertex AI.