Agent frameworks and SDKs
Libraries for building your own agents: model calls, tools, memory, orchestration and multi-agent patterns.
By weekly package downloads, the leaders are LangChain (44.3M), AI SDK (34.5M), Claude Agent SDK (21.5M). By GitHub stars, the leaders are LangChain (147K), MetaGPT (70.8K), AutoGen (61.3K). The fastest grower over the last 30 days is AgentKit by Inngest, with downloads up 74%. As of Oct 6, 2026.
| # | Agent | Pulse | Downloads 7d | 7d | 30d | VS Code installs | Stars | Latest release | Price | Last 90 days | |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 4 |
LangChainLangChain |
77 | 44.3M | up 1.9% | down 22.0% | — | 147K+45/day | 1.4.38d ago | Free (OSS) | ||
| 10 |
LangGraphLangChain |
69 | 16.1M | up 11.4% | down 21.8% | — | 42.7K+48/day | 1.2.13yesterday | Free (OSS) | ||
| 11 |
AI SDKVercel |
67 | 34.5M | up 13.5% | up 20.9% | — | 27.1K+18/day | 6.0.301yesterday | Free (OSS) | ||
| 13 | 63 | 8.7M | up 0.6% | up 2.0% | — | 8,679+23/day | 1.58.0yesterday | Free (OSS) | |||
| 17 |
Claude Agent SDKAnthropic |
59 | 21.5M | up 14.6% | up 0.6% | — | 8,219+5/day | 0.2.1636d ago | Free (OSS) | ||
| 21 |
MastraMastra |
58 | 2.2M | up 11.1% | up 19.1% | — | 28.6K+27/day | 1.74.0yesterday | Free (OSS) | ||
| 23 |
Deep AgentsLangChain |
57 | 2M | up 13.5% | up 5.0% | — | 30K+19/day | 0.7.22yesterday | Free (OSS) | ||
| 25 |
Cloudflare Agents SDKCloudflare |
56 | 2.3M | up 11.0% | up 23.9% | — | 5,778+18/day | 0.26.04d ago | Free (OSS) | ||
| 29 |
OpenAI Agents SDKOpenAI |
54 | 5.5M | up 7.7% | down 46.4% | — | 29.8K+16/day | 0.23.14d ago | Free (OSS) | ||
| 30 |
DSPyStanford NLP |
53 | 1.4M | up 4.8% | down 13.9% | — | 38.5K+16/day | 3.4.011d ago | Free (OSS) | ||
| 32 |
AgnoAgno |
52 | 506K | up 12.7% | down 15.2% | — | 42.6K+31/day | 3.1.14d ago | Free (OSS) | ||
| 33 |
CrewAICrewAI |
52 | 611K | up 7.2% | — | — | 59.4K+37/day | 1.15.238d ago | Free (OSS) | ||
| 49 | 47 | 2.6M | down 9.6% | down 42.6% | — | 21.7K+8/day | 2.11.04d ago | Free (OSS) | |||
| 50 |
LlamaIndexLlamaIndex |
46 | 848K | up 2.6% | down 48.0% | — | 52.4K+12/day | 0.14.2515d ago | Free (OSS) | ||
| 56 |
Pydantic AIPydantic |
44 | 1.4M | up 6.6% | down 48.8% | — | 20.4K+26/day | 2.54.03d ago | Free (OSS) | ||
| 57 |
Haystackdeepset |
43 | 160K | up 12.2% | down 24.0% | — | 26.7K+9/day | 3.3.05d ago | Free (OSS) | ||
| 63 |
VoltAgentVoltAgent |
42 | 55.2K | up 83.3% | up 41.1% | — | 10.7K+6/day | 2.11.08d ago | Free (OSS) | ||
| 69 |
AutoGenMicrosoft |
39 | 89K | up 0.4% | down 54.7% | — | 61.3K+14/day | 0.7.512mo ago | Free (OSS) | ||
| 71 |
Semantic KernelMicrosoft |
39 | 79.1K | 0.0% | — | — | 28.6K+5/day | 1.80.133d ago | Free (OSS) | ||
| 76 |
smolagentsHugging Face |
36 | 108K | up 5.9% | down 25.0% | — | 29.7K+19/day | 1.26.04mo ago | Free (OSS) | ||
| 82 |
Microsoft Agent FrameworkMicrosoft |
35 | 55.5K | up 5.8% | — | — | 14K+15/day | 1.20.04d ago | Free (OSS) | ||
| 95 |
AgentKit by InngestInngest |
32 | 31.1K | down 59.2% | up 73.8% | — | 939 | 0.13.210mo ago | Free (OSS) | ||
| 100 |
LettaLetta |
30 | 12.4K | up 27.5% | — | — | 25K+12/day | 0.16.84mo ago | Free (OSS) | ||
| 101 |
CAMELCAMEL-AI |
29 | 16.3K | up 33.5% | down 30.9% | — | 17.8K+4/day | 0.2.906mo ago | Free (OSS) | ||
| 106 |
MetaGPTFoundationAgents |
26 | 855 | up 2.2% | — | — | 70.8K+15/day | 0.8.219mo ago | Free (OSS) | ||
| 108 |
AG2AG2 |
25 | 31.9K | down 20.8% | down 44.5% | — | 4,975+2/day | 1.1.23d ago | Free (OSS) | ||
| 123 |
BeeAI FrameworkBeeAI (Linux Foundation) |
18 | 10.5K | up 42.8% | down 40.9% | — | 3,426 | 0.1.858d ago | Free (OSS) | ||
| No agents match that filter. | |||||||||||
Ranks are overall positions by Pulse Score. Click a column to sort by what matters to you. Methodology.
What a framework gives you
At its core an agent is a loop: call a model, run the tools it asks for, feed the results back, repeat. A framework supplies the parts around that loop: tool definitions, state and memory, retries, streaming, tracing, human approval steps and ways to coordinate several agents.
The main families
- Vendor SDKs from model providers are thin, track new model features first and work best with that vendor's models.
- Orchestration frameworks model the agent as a graph or workflow with explicit state, which helps with long-running and resumable work.
- Multi-agent frameworks organise work as roles that hand tasks to each other.
- Typed, minimal libraries stay close to plain code and suit teams that want few abstractions.
Reading the numbers
Downloads are a strong signal for libraries, but they count installs in CI and as dependencies of other packages, so a library that many others depend on is downloaded far more often than it is chosen. Use the 30-day change and release cadence to see which projects are gaining ground.
Head-to-head comparisons
Related guides
AI Agent vs Chatbot vs Agentic AI: What Is the Difference?
Plain-language definitions of AI agents, chatbots, agentic AI, LLMs, workflows and skills, how they relate, and how to tell which one you are looking at, with real examples.
Best AI Agents in 2026: Live Rankings by Category
The leading AI agents for coding, automation, research, browsing, app building, customer support, voice and sales, ranked by public usage data that refreshes daily, with prices and free options.
How to Build an AI Agent: No-Code Builders, Frameworks and SDKs
A practical guide to building your first AI agent: choosing between no-code builders, open-source frameworks and model vendors' SDKs, the core agent loop, tools, guardrails and testing, with live usage data for each option.
What Is an AI Agent? A Practical Explanation With Real Examples
What makes software an AI agent, how agents differ from chatbots and plain language models, the main types in use today and how to judge one.
AI Agent Statistics 2026: Live Usage, Growth and Activity Data
Current statistics on AI agents from public data: package downloads, editor installs, GitHub stars, pull requests opened by coding agents, release pace and benchmark results. Updated daily.
Questions about agent frameworks and SDKs
What is the most downloaded AI agent framework?
LangChain, with 44.3M downloads in the last 7 days, then AI SDK (34.5M) and Claude Agent SDK (21.5M). Counts include installs as a dependency of other packages.
Do I need a framework to build an AI agent?
No. A basic agent is a short loop around a model API with tool calling. Frameworks pay off when you need persistence, human approval steps, tracing, or coordination between several agents. A step-by-step guide is in [how to build an AI agent](/guides/how-to-build-an-ai-agent/).