AgentStackWatch
⌕Subscribe

Reference

The Tracker

43 products refreshed nightly from GitHub, PyPI and npm. ✓ marks entries checked against sources; anything unchecked reads unknown.

Personal agentsManaged agent platformsDev frameworks & SDKsResearch & benchmarks

Personal agents

Consumer and self-hosted assistants that act on your behalf.

ProductPricingVersionStarsDownloads / wk26-week trend
Grok Bot✓xAI Included in X Premium ($30/mo), Super ($100/mo), Ultra ($200/mo); not in Plus. Team seats: Premium $30, Super $100, Ultra $200 per seat/mo —— — —
Muse from Meta✓Meta unknown —— — —
OpenClaw✓OpenClaw Foundation (501(c)(3)) Free; no paid tier or hosted service 2026.10.4392.2k 3.9M
ChatGPT dots✓OpenAI First dot included in Pro, Business Premium, Enterprise (rolling out) —— — —
Hermes Agent✓Nous Research Free (MIT); optional Nous Portal subscription 0.22.2254.5k — —
Manus✓Manus AI Free $0; Pro from $20/mo (4,000 credits) or $40/mo; Team from $20/seat/mo; Enterprise custom —— — —
Lindy✓Lindy From $29.99/user/mo (3,000 credits); Plus, Pro, Max tiers; Enterprise custom; $50 free credits for 7 days —— — —
Perplexity Comet✓Perplexity Free; Comet Plus $5/mo (included in Perplexity Pro/Max) —— — —
Gemini Agent✓Google Google AI Ultra ($99.99/mo 5x or $199.99/mo 20x); US, English only —— — —
Claude (app)✓Anthropic Free; Pro $20/mo ($17/mo annual); Max 5x $100/mo; Max 20x $200/mo —— — —

Managed agent platforms

Hosted runtimes, agent APIs and sandboxes.

ProductPricingVersionStarsDownloads / wk26-week trend
Claude Managed Agents✓Anthropic Model token rates + $0.08 per active session-hour; web search $10/1k —— — —
OpenAI Agents API✓OpenAI No platform fee; tokens + tool/container rates —— — —
Vertex AI Agent Engine✓Google Cloud Agent Compute $0.085/vCPU-hr (50 vCPU-hr/mo free); memory $0.009/GiB-hr (100 GiB-hr free); since 2026-06-17 —— — —
Amazon Bedrock AgentCore✓AWS Runtime microVM V1 $0.0895/vCPU-hr + $0.00945/GB-hr; V2 $0.1276/vCPU-hr + $0.0169/GB-hr; per-second —— — —
Microsoft Foundry Agent Service✓Microsoft No charge for prompt/workflow agents (pay tokens + tools); hosted agents billed per vCPU/GiB-hour —— — —
Relevance AI✓Relevance AI Pro $19/mo annual or $29 monthly (2,500 actions); Team $234/mo annual or $349 monthly; Enterprise custom; free plan retired —— — —
Letta Cloud✓Letta Free (3 agents, BYOK); Pro $20/mo (20 agents); API plan $20/mo + $0.10/active agent/mo + usage; Enterprise custom —— — —
LangSmith Deployment (formerly LangGraph Platform)✓LangChain Plus plan includes 1 free small serverless deployment; extra deployments billed by resources —— — —
CrewAI AMP✓CrewAI Basic free (50 workflow executions/mo); Enterprise custom —— — —
E2B✓E2B Hobby free ($100 one-time credit); Pro $150/mo + usage; Enterprise custom 2.56.014.4k 1.7M
Modal✓Modal Starter $0/mo with $30/mo compute credits; per-second usage after —— — —
Daytona✓Daytona Pay as you go: $0.0504/vCPU-h + $0.0162/GiB-h; $200 free compute 0.190.071.5k — —

Dev frameworks & SDKs

Libraries for building agents in code.

ProductPricingVersionStarsDownloads / wk26-week trend
LangGraph✓LangChain Free / open source 1.2.1643.4k 12.2M
Pydantic AI✓Pydantic Free / open source 2.57.020.7k 1.4M
OpenAI Agents SDK✓OpenAI Free / open source 0.24.030.1k 3.0M
Claude Agent SDK✓Anthropic Free / open source 0.2.1688.3k 7.6M
Google ADK✓Google Free / open source 2.12.021.9k 2.2M
CrewAI✓CrewAI Free / open source 1.15.2859.8k 633.5k
Mastra✓Mastra Free / open source 1.78.028.9k 2.2M
Vercel AI SDK✓Vercel Free / open source 7.0.14327.4k 31.0M
LlamaIndex✓LlamaIndex Free / open source 0.14.2552.5k 664.2k
Haystack✓deepset Free / open source 3.4.026.8k 131.3k
Letta✓Letta Free / open source 0.16.825.2k — —
DSPy✓Stanford NLP Free / open source 3.4.138.7k 1.2M
smolagents✓Hugging Face Free / open source 1.26.029.9k 123.1k
Browser Use✓Browser Use Free / open source 0.13.11118.2k 1.7M
OpenHands✓OpenHands (formerly All Hands AI) Free / open source 1.27.191.1k — —
Goose✓unknown Free / open source 1.55.055.4k — —

Research & benchmarks

Leaderboards and benchmarks used to evaluate agents.

ProductPricingVersionStarsDownloads / wk26-week trend
HAL (Holistic Agent Leaderboard)✓Princeton SAgE unknown —— — —
SWE-bench✓SWE-bench team unknown —6.0k — —
GAIA✓Meta & Hugging Face researchers unknown —— — —
τ-bench✓Sierra unknown —1.5k — —
OSWorld✓unknown unknown 0.1.163.2k — —