ModelCompare
Compare

AI coding agent usage

What was the most used AI coding agent on ?

OpenClaw from OpenClaw, with a usage score of 52 out of 100 across 3 of 4 sources. Claude Code was next at 42.

Usage score over time, week by week

Usage ranking

AI coding agents by usage on July 1, 2026

This ranks how much each agent is used and followed, not how good it is: a popular agent can be a worse fit for your work than a small one. Click an agent to see where its numbers come from. Each source's column shows the agent's scaled score (the leader is 100) over the raw number. 4 sources had data on this day. The score averages an agent's scaled values over the sources that can measure it (at least three).

  1. 1
    OpenClaw
    OpenClaw · open source
    Usage score 52
    • npm 28 · 3.1M
    • OpenRouter 29 · 263B
    • GitHub 100 · 376k
    • npm: 283,090,596 downloads of openclaw in the 7 days to Jul 1, 2026. The leader here is Codex CLI with 11M, so OpenClaw scores 3.1M ÷ 11M × 100 = 28. Source
    • OpenRouter: 29263B tokens sent through OpenRouter by OpenClaw in the week to May 1, 2026. The leader here is Cline with 918B, so OpenClaw scores 263B ÷ 918B × 100 = 29. Source
    • GitHub: 100376,460 stars on github.com/openclaw/openclaw on Jun 3, 2026. That's the most of any agent in this source, so it scores 100. Source

    Not counted: VS Code, because it has no VS Code extension.

    Usage score: (28 + 29 + 100) ÷ 3 = 52. OpenClaw website

  2. 2
    Claude Code
    Anthropic · closed source
    Usage score 42
    • npm 94 · 10M
    • OpenRouter 14 · 126B
    • GitHub 34 · 129k
    • VS Code 25 · 19M
    • npm: 9410,180,497 downloads of @anthropic-ai/claude-code in the 7 days to Jul 1, 2026. The leader here is Codex CLI with 11M, so Claude Code scores 10M ÷ 11M × 100 = 94. Source
    • OpenRouter: 14126B tokens sent through OpenRouter by Claude Code in the week to May 1, 2026. The leader here is Cline with 918B, so Claude Code scores 126B ÷ 918B × 100 = 14. Source
    • GitHub: 34129,226 stars on github.com/anthropics/claude-code on Jun 1, 2026. The leader here is OpenClaw with 376k, so Claude Code scores 129k ÷ 376k × 100 = 34. Source
    • VS Code: 2519,278,058 installs of the anthropic.claude-code extension on Jul 1, 2026. The leader here is GitHub Copilot with 76M, so Claude Code scores 19M ÷ 76M × 100 = 25. Source

    Usage score: (94 + 14 + 34 + 25) ÷ 4 = 42. Claude Code website

  3. 3
    GitHub Copilot
    GitHub · closed source
    Usage score 36
    • npm 9 · 944k
    • VS Code 100 · 76M
    • npm: 9944,379 downloads of @github/copilot in the 7 days to Jul 1, 2026. The leader here is Codex CLI with 11M, so GitHub Copilot scores 944k ÷ 11M × 100 = 9. Source
    • VS Code: 10076,122,820 installs of the GitHub.copilot-chat extension on Jul 1, 2026. That's the most of any agent in this source, so it scores 100. Source

    Not counted: OpenRouter, because it doesn't show up in OpenRouter's app rankings; GitHub, because it has no public GitHub repo.

    Usage score: (9 + 100) ÷ 3 = 36, divided by 3 rather than 2 because every agent's average counts at least 3 sources. GitHub Copilot website

  4. 4
    Codex CLI
    OpenAI · open source
    Usage score 35
    • npm 100 · 11M
    • OpenRouter 2 · 18B
    • GitHub 25 · 95k
    • VS Code 13 · 9.7M
    • npm: 10010,846,526 downloads of @openai/codex in the 7 days to Jul 1, 2026. That's the most of any agent in this source, so it scores 100. Source
    • OpenRouter: 218B tokens sent through OpenRouter by Codex CLI in the week to May 1, 2026. The leader here is Cline with 918B, so Codex CLI scores 18B ÷ 918B × 100 = 2. Source
    • GitHub: 2594,956 stars on github.com/openai/codex on Jul 2, 2026. The leader here is OpenClaw with 376k, so Codex CLI scores 95k ÷ 376k × 100 = 25. Source
    • VS Code: 139,651,527 installs of the openai.chatgpt extension on Jun 1, 2026. The leader here is GitHub Copilot with 76M, so Codex CLI scores 9.7M ÷ 76M × 100 = 13. Source

    Usage score: (100 + 2 + 25 + 13) ÷ 4 = 35. Codex CLI website

  5. 5
    Cline
    Cline · open source
    Usage score 31
    • npm 0 · 49k
    • OpenRouter 100 · 918B
    • GitHub 17 · 64k
    • VS Code 6 · 4.2M
    • npm: 048,596 downloads of cline in the 7 days to Jul 1, 2026. The leader here is Codex CLI with 11M, so Cline scores 49k ÷ 11M × 100 = 0. Source
    • OpenRouter: 100918B tokens sent through OpenRouter by Cline in the week to May 1, 2026. That's the most of any agent in this source, so it scores 100. Source
    • GitHub: 1764,144 stars on github.com/cline/cline on Jul 1, 2026. The leader here is OpenClaw with 376k, so Cline scores 64k ÷ 376k × 100 = 17. Source
    • VS Code: 64,205,261 installs of the saoudrizwan.claude-dev extension on Jun 5, 2026. The leader here is GitHub Copilot with 76M, so Cline scores 4.2M ÷ 76M × 100 = 6. Source

    Usage score: (0 + 100 + 17 + 6) ÷ 4 = 31. Cline website

  6. 6
    Hermes Agent
    Nous Research · open source
    Usage score 25
    • OpenRouter 21 · 196B
    • GitHub 55 · 207k
    • OpenRouter: 21196B tokens sent through OpenRouter by Hermes Agent in the week to May 1, 2026. The leader here is Cline with 918B, so Hermes Agent scores 196B ÷ 918B × 100 = 21. Source
    • GitHub: 55206,768 stars on github.com/NousResearch/hermes-agent on Jul 1, 2026. The leader here is OpenClaw with 376k, so Hermes Agent scores 207k ÷ 376k × 100 = 55. Source

    Not counted: npm, because it has no command-line package on npm; VS Code, because it has no VS Code extension.

    Usage score: (21 + 55) ÷ 3 = 25, divided by 3 rather than 2 because every agent's average counts at least 3 sources. Hermes Agent website

  7. 7
    Roo Code
    Roo Code · open source
    Usage score 22
    • OpenRouter 59 · 539B
    • GitHub 6 · 24k
    • VS Code 2 · 1.7M
    • OpenRouter: 59539B tokens sent through OpenRouter by Roo Code in the week to May 1, 2026. The leader here is Cline with 918B, so Roo Code scores 539B ÷ 918B × 100 = 59. Source
    • GitHub: 624,264 stars on github.com/RooCodeInc/Roo-Code on Jun 23, 2026. The leader here is OpenClaw with 376k, so Roo Code scores 24k ÷ 376k × 100 = 6. Source
    • VS Code: 21,722,382 installs of the RooVeterinaryInc.roo-cline extension on Jun 10, 2026. The leader here is GitHub Copilot with 76M, so Roo Code scores 1.7M ÷ 76M × 100 = 2. Source

    Not counted: npm, because it has no command-line package on npm.

    Usage score: (59 + 6 + 2) ÷ 3 = 22. Roo Code website

  8. 8
    OpenCode
    Anomaly · open source
    Usage score 21
    • npm 14 · 1.5M
    • GitHub 48 · 181k
    • VS Code 1 · 674k
    • npm: 141,522,002 downloads of opencode-ai in the 7 days to Jul 1, 2026. The leader here is Codex CLI with 11M, so OpenCode scores 1.5M ÷ 11M × 100 = 14. Source
    • GitHub: 48181,008 stars on github.com/anomalyco/opencode on Jul 1, 2026. The leader here is OpenClaw with 376k, so OpenCode scores 181k ÷ 376k × 100 = 48. Source
    • VS Code: 1673,719 installs of the sst-dev.opencode extension on Jun 8, 2026. The leader here is GitHub Copilot with 76M, so OpenCode scores 674k ÷ 76M × 100 = 1. Source

    Not counted: OpenRouter, because it doesn't show up in OpenRouter's app rankings.

    Usage score: (14 + 48 + 1) ÷ 3 = 21. OpenCode website

  9. 9
    OpenHands
    All Hands AI · open source
    Usage score 12
    • OpenRouter 17 · 152B
    • GitHub 20 · 76k
    • OpenRouter: 17152B tokens sent through OpenRouter by OpenHands in the week to May 1, 2026. The leader here is Cline with 918B, so OpenHands scores 152B ÷ 918B × 100 = 17. Source
    • GitHub: 2076,145 stars on github.com/OpenHands/OpenHands on Jun 7, 2026. The leader here is OpenClaw with 376k, so OpenHands scores 76k ÷ 376k × 100 = 20. Source

    Not counted: npm, because it has no command-line package on npm; VS Code, because it has no VS Code extension.

    Usage score: (17 + 20) ÷ 3 = 12, divided by 3 rather than 2 because every agent's average counts at least 3 sources. OpenHands website

  10. 10
    Gemini CLI
    Google · open source
    Usage score 11
    • npm 4 · 433k
    • GitHub 28 · 105k
    • npm: 4433,141 downloads of @google/gemini-cli in the 7 days to Jul 1, 2026. The leader here is Codex CLI with 11M, so Gemini CLI scores 433k ÷ 11M × 100 = 4. Source
    • GitHub: 28104,903 stars on github.com/google-gemini/gemini-cli on Jun 3, 2026. The leader here is OpenClaw with 376k, so Gemini CLI scores 105k ÷ 376k × 100 = 28. Source

    Not counted: OpenRouter, because it doesn't show up in OpenRouter's app rankings; VS Code, because there's no snapshot of its extension page from around that day.

    Usage score: (4 + 28) ÷ 3 = 11, divided by 3 rather than 2 because every agent's average counts at least 3 sources. Gemini CLI website

  11. 11
    Kilo Code
    Kilo · open source
    Usage score 7
    • npm 0 · 52k
    • OpenRouter 21 · 190B
    • GitHub 5 · 20k
    • VS Code 2 · 1.2M
    • npm: 052,058 downloads of @kilocode/cli in the 7 days to Jul 1, 2026. The leader here is Codex CLI with 11M, so Kilo Code scores 52k ÷ 11M × 100 = 0. Source
    • OpenRouter: 21190B tokens sent through OpenRouter by Kilo Code in the week to May 1, 2026. The leader here is Cline with 918B, so Kilo Code scores 190B ÷ 918B × 100 = 21. Source
    • GitHub: 519,822 stars on github.com/Kilo-Org/kilocode on Jun 4, 2026. The leader here is OpenClaw with 376k, so Kilo Code scores 20k ÷ 376k × 100 = 5. Source
    • VS Code: 21,183,475 installs of the kilocode.Kilo-Code extension on Jun 8, 2026. The leader here is GitHub Copilot with 76M, so Kilo Code scores 1.2M ÷ 76M × 100 = 2. Source

    Usage score: (0 + 21 + 5 + 2) ÷ 4 = 7. Kilo Code website

  12. 12
    Qwen Code
    Alibaba · open source
    Usage score 5
    • npm 1 · 118k
    • OpenRouter 8 · 74B
    • GitHub 7 · 26k
    • npm: 1117,716 downloads of @qwen-code/qwen-code in the 7 days to Jul 1, 2026. The leader here is Codex CLI with 11M, so Qwen Code scores 118k ÷ 11M × 100 = 1. Source
    • OpenRouter: 874B tokens sent through OpenRouter by Qwen Code in the week to May 1, 2026. The leader here is Cline with 918B, so Qwen Code scores 74B ÷ 918B × 100 = 8. Source
    • GitHub: 725,693 stars on github.com/QwenLM/qwen-code on Jul 1, 2026. The leader here is OpenClaw with 376k, so Qwen Code scores 26k ÷ 376k × 100 = 7. Source

    Not counted: VS Code, because it has no VS Code extension.

    Usage score: (1 + 8 + 7) ÷ 3 = 5. Qwen Code website

  13. 13
    Aider
    Aider · open source
    Usage score 5
    • OpenRouter 1 · 12B
    • GitHub 12 · 47k
    • OpenRouter: 112B tokens sent through OpenRouter by Aider in the week to May 1, 2026. The leader here is Cline with 918B, so Aider scores 12B ÷ 918B × 100 = 1. Source
    • GitHub: 1246,890 stars on github.com/Aider-AI/aider on Jul 1, 2026. The leader here is OpenClaw with 376k, so Aider scores 47k ÷ 376k × 100 = 12. Source

    Not counted: npm, because it has no command-line package on npm; VS Code, because it has no VS Code extension.

    Usage score: (1 + 12) ÷ 3 = 5, divided by 3 rather than 2 because every agent's average counts at least 3 sources. Aider website

  14. 14
    Crush
    Charm · open source
    Usage score 2
    • npm 0 · 3.4k
    • GitHub 7 · 26k
    • npm: 03,429 downloads of @charmland/crush in the 7 days to Jul 1, 2026. The leader here is Codex CLI with 11M, so Crush scores 3.4k ÷ 11M × 100 = 0. Source
    • GitHub: 725,963 stars on github.com/charmbracelet/crush on Jul 2, 2026. The leader here is OpenClaw with 376k, so Crush scores 26k ÷ 376k × 100 = 7. Source

    Not counted: OpenRouter, because it doesn't show up in OpenRouter's app rankings; VS Code, because it has no VS Code extension.

    Usage score: (0 + 7) ÷ 3 = 2, divided by 3 rather than 2 because every agent's average counts at least 3 sources. Crush website

  15. 15
    pi
    Mario Zechner · open source
    Usage score 1
    • npm 0 · 7.6k
    • OpenRouter 4 · 35B
    • npm: 07,588 downloads of @mariozechner/pi-coding-agent in the 7 days to Jul 1, 2026. The leader here is Codex CLI with 11M, so pi scores 7.6k ÷ 11M × 100 = 0. Source
    • OpenRouter: 435B tokens sent through OpenRouter by pi in the week to May 1, 2026. The leader here is Cline with 918B, so pi scores 35B ÷ 918B × 100 = 4. Source

    Not counted: GitHub, because there's no snapshot of its repo from around that day; VS Code, because it has no VS Code extension.

    Usage score: (0 + 4) ÷ 3 = 1, divided by 3 rather than 2 because every agent's average counts at least 3 sources. pi website

  16. 16
    Letta Code
    Letta · open source
    Usage score 0
    • npm 0 · 7.8k
    • OpenRouter 0 · 0
    • GitHub 1 · 2.6k
    • npm: 07,830 downloads of @letta-ai/letta-code in the 7 days to Jul 1, 2026. The leader here is Codex CLI with 11M, so Letta Code scores 7.8k ÷ 11M × 100 = 0. Source
    • OpenRouter: 00 tokens sent through OpenRouter by Letta Code in the week to May 1, 2026. The leader here is Cline with 918B, so Letta Code scores 0 ÷ 918B × 100 = 0. Source
    • GitHub: 12,553 stars on github.com/letta-ai/letta-code on May 24, 2026. The leader here is OpenClaw with 376k, so Letta Code scores 2.6k ÷ 376k × 100 = 1. Source

    Not counted: VS Code, because it has no VS Code extension.

    Usage score: (0 + 0 + 1) ÷ 3 = 0. Letta Code website

Not ranked that day, only one source has them: Auggie (npm, 34k), Amp (npm, 16k), Droid (npm, 1.8k).

Method

How the usage score is made

No single public number measures how much each AI coding agent is used, so we blend the ones that exist. In each source the leader scores 100 and the rest are scaled to it. An agent's score is the average over the sources that can measure it. Sources that can't see an agent at all (Cursor has no npm package) are left out of its average, but the average always counts at least three sources, so an agent seen by only one or two can't outrank one that leads everywhere. An agent needs two sources to be ranked.

  • npmCLI downloads, 7 days · data from Jul 1, 2026

    Downloads of each agent's command-line package in the 7 days to the date, with one-off bursts (mirrors, bots) capped. Downloads count installs and CI runs, not people, and agents with their own installer are undercounted. Source

  • OpenRouterTokens via OpenRouter, 7 days · data from May 1, 2026

    Tokens each app sent through OpenRouter in a week. It only sees traffic routed through OpenRouter, so agents that mostly call their own lab's API (Claude Code, Codex) look smaller than they are, and open-source agents that default to OpenRouter look bigger. Source

  • GitHubStars · data from Jul 2, 2026

    Stars on the agent's GitHub repo, from monthly snapshots. Stars measure interest, and build up over time. Source

  • VS CodeExtension installs, all time · data from Jul 1, 2026

    All-time installs of the agent's VS Code extension, from monthly snapshots. Cursor and Windsurf install from Open VSX instead, which isn't counted. Source

Browse

The most used AI coding agent, month by month

2026
2025

Pick the model behind your agent

Most of these agents let you choose the model. Send one prompt to up to six models and see which answer you'd actually use, with what each one cost.

Compare models

Data through October 21, 2026 for 20 agents. npm downloads from the npm registry; OpenRouter app rankings from OpenRouter (CC BY 4.0); GitHub stars and VS Code installs from monthly snapshots in the Internet Archive and the live APIs; survey figures as published by Stack Overflow and JetBrains.