ModelCompare
Compare

AI coding agent usage

What was the most used AI coding agent on ?

Claude Code from Anthropic, with a usage score of 48 out of 100 across 3 of 3 sources. Gemini CLI was next at 35.

Usage score over time, week by week

Usage ranking

AI coding agents by usage on October 1, 2025

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. 3 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
    Claude Code
    Anthropic · closed source
    Usage score 48
    • npm 100 · 5.8M
    • GitHub 41 · 32k
    • VS Code 2 · 752k
    • npm: 1005,805,186 downloads of @anthropic-ai/claude-code in the 7 days to Oct 1, 2025. That's the most of any agent in this source, so it scores 100. Source
    • GitHub: 4132,130 stars on github.com/anthropics/claude-code on Sep 1, 2025. The leader here is Gemini CLI with 78k, so Claude Code scores 32k ÷ 78k × 100 = 41. Source
    • VS Code: 2751,944 installs of the anthropic.claude-code extension on Sep 16, 2025. The leader here is GitHub Copilot with 40M, so Claude Code scores 752k ÷ 40M × 100 = 2. Source

    Usage score: (100 + 41 + 2) ÷ 3 = 48. Claude Code website

  2. 2
    Gemini CLI
    Google · open source
    Usage score 35
    • npm 4 · 232k
    • GitHub 100 · 78k
    • VS Code 0 · 91k
    • npm: 4231,684 downloads of @google/gemini-cli in the 7 days to Oct 1, 2025. The leader here is Claude Code with 5.8M, so Gemini CLI scores 232k ÷ 5.8M × 100 = 4. Source
    • GitHub: 10077,576 stars on github.com/google-gemini/gemini-cli on Oct 1, 2025. That's the most of any agent in this source, so it scores 100. Source
    • VS Code: 091,335 installs of the Google.gemini-cli-vscode-ide-companion extension on Sep 3, 2025. The leader here is GitHub Copilot with 40M, so Gemini CLI scores 91k ÷ 40M × 100 = 0. Source

    Usage score: (4 + 100 + 0) ÷ 3 = 35. Gemini CLI website

  3. 3
    GitHub Copilot
    GitHub · closed source
    Usage score 34
    • npm 1 · 57k
    • VS Code 100 · 40M
    • npm: 157,468 downloads of @github/copilot in the 7 days to Oct 1, 2025. The leader here is Claude Code with 5.8M, so GitHub Copilot scores 57k ÷ 5.8M × 100 = 1. Source
    • VS Code: 10040,007,219 installs of the GitHub.copilot-chat extension on Sep 8, 2025. That's the most of any agent in this source, so it scores 100. Source

    Not counted: GitHub, because it has no public GitHub repo.

    Usage score: (1 + 100) ÷ 3 = 34, divided by 3 rather than 2 because every agent's average counts at least 3 sources. GitHub Copilot website

  4. 4
    Cline
    Cline · open source
    Usage score 24
    • npm 0 · 2.7k
    • GitHub 66 · 51k
    • VS Code 6 · 2.3M
    • npm: 02,749 downloads of cline in the 7 days to Oct 1, 2025. The leader here is Claude Code with 5.8M, so Cline scores 2.7k ÷ 5.8M × 100 = 0. Source
    • GitHub: 6650,972 stars on github.com/cline/cline on Oct 1, 2025. The leader here is Gemini CLI with 78k, so Cline scores 51k ÷ 78k × 100 = 66. Source
    • VS Code: 62,345,153 installs of the saoudrizwan.claude-dev extension on Oct 1, 2025. The leader here is GitHub Copilot with 40M, so Cline scores 2.3M ÷ 40M × 100 = 6. Source

    Usage score: (0 + 66 + 6) ÷ 3 = 24. Cline website

  5. 5
    Codex CLI
    OpenAI · open source
    Usage score 21
    • npm 3 · 190k
    • GitHub 59 · 46k
    • VS Code 2 · 638k
    • npm: 3190,451 downloads of @openai/codex in the 7 days to Oct 1, 2025. The leader here is Claude Code with 5.8M, so Codex CLI scores 190k ÷ 5.8M × 100 = 3. Source
    • GitHub: 5945,626 stars on github.com/openai/codex on Oct 2, 2025. The leader here is Gemini CLI with 78k, so Codex CLI scores 46k ÷ 78k × 100 = 59. Source
    • VS Code: 2638,141 installs of the openai.chatgpt extension on Sep 6, 2025. The leader here is GitHub Copilot with 40M, so Codex CLI scores 638k ÷ 40M × 100 = 2. Source

    Usage score: (3 + 59 + 2) ÷ 3 = 21. Codex CLI website

  6. 6
    Roo Code
    Roo Code · open source
    Usage score 9
    • GitHub 26 · 20k
    • VS Code 2 · 815k
    • GitHub: 2620,016 stars on github.com/RooCodeInc/Roo-Code on Oct 2, 2025. The leader here is Gemini CLI with 78k, so Roo Code scores 20k ÷ 78k × 100 = 26. Source
    • VS Code: 2815,431 installs of the RooVeterinaryInc.roo-cline extension on Sep 4, 2025. The leader here is GitHub Copilot with 40M, so Roo Code scores 815k ÷ 40M × 100 = 2. Source

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

    Usage score: (26 + 2) ÷ 3 = 9, divided by 3 rather than 2 because every agent's average counts at least 3 sources. Roo Code website

  7. 7
    Qwen Code
    Alibaba · open source
    Usage score 5
    • npm 1 · 35k
    • GitHub 15 · 12k
    • npm: 135,061 downloads of @qwen-code/qwen-code in the 7 days to Oct 1, 2025. The leader here is Claude Code with 5.8M, so Qwen Code scores 35k ÷ 5.8M × 100 = 1. Source
    • GitHub: 1511,837 stars on github.com/QwenLM/qwen-code on Sep 1, 2025. The leader here is Gemini CLI with 78k, so Qwen Code scores 12k ÷ 78k × 100 = 15. Source

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

    Usage score: (1 + 15) ÷ 3 = 5, divided by 3 rather than 2 because every agent's average counts at least 3 sources. Qwen Code website

  8. 8
    Crush
    Charm · open source
    Usage score 5
    • npm 0 · 2.3k
    • GitHub 15 · 12k
    • npm: 02,274 downloads of @charmland/crush in the 7 days to Oct 1, 2025. The leader here is Claude Code with 5.8M, so Crush scores 2.3k ÷ 5.8M × 100 = 0. Source
    • GitHub: 1511,997 stars on github.com/charmbracelet/crush on Sep 5, 2025. The leader here is Gemini CLI with 78k, so Crush scores 12k ÷ 78k × 100 = 15. Source

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

    Usage score: (0 + 15) ÷ 3 = 5, divided by 3 rather than 2 because every agent's average counts at least 3 sources. Crush website

  9. 9
    Kilo Code
    Kilo · open source
    Usage score 5
    • npm 0 · 0
    • GitHub 14 · 11k
    • VS Code 1 · 350k
    • npm: 00 downloads of @kilocode/cli in the 7 days to Oct 1, 2025. The leader here is Claude Code with 5.8M, so Kilo Code scores 0 ÷ 5.8M × 100 = 0. Source
    • GitHub: 1410,656 stars on github.com/Kilo-Org/kilocode on Oct 1, 2025. The leader here is Gemini CLI with 78k, so Kilo Code scores 11k ÷ 78k × 100 = 14. Source
    • VS Code: 1350,152 installs of the kilocode.Kilo-Code extension on Sep 27, 2025. The leader here is GitHub Copilot with 40M, so Kilo Code scores 350k ÷ 40M × 100 = 1. Source

    Usage score: (0 + 14 + 1) ÷ 3 = 5. Kilo Code website

  10. 10
    Auggie
    Augment Code · closed source
    Usage score 1
    • npm 0 · 15k
    • VS Code 1 · 536k
    • npm: 014,899 downloads of @augmentcode/auggie in the 7 days to Oct 1, 2025. The leader here is Claude Code with 5.8M, so Auggie scores 15k ÷ 5.8M × 100 = 0. Source
    • VS Code: 1536,481 installs of the augment.vscode-augment extension on Sep 15, 2025. The leader here is GitHub Copilot with 40M, so Auggie scores 536k ÷ 40M × 100 = 1. Source

    Not counted: GitHub, because it has no public GitHub repo.

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

  11. 11
    OpenCode
    Anomaly · open source
    Usage score 0
    • npm 1 · 46k
    • VS Code 0 · 1.8k
    • npm: 146,436 downloads of opencode-ai in the 7 days to Oct 1, 2025. The leader here is Claude Code with 5.8M, so OpenCode scores 46k ÷ 5.8M × 100 = 1. Source
    • VS Code: 01,847 installs of the sst-dev.opencode extension on Aug 13, 2025. The leader here is GitHub Copilot with 40M, so OpenCode scores 1.8k ÷ 40M × 100 = 0. Source

    Not counted: GitHub, because there's no snapshot of its repo from around that day.

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

  12. 12
    Amp
    Sourcegraph · closed source
    Usage score 0
    • npm 0 · 14k
    • VS Code 0 · 39k
    • npm: 014,218 downloads of @sourcegraph/amp in the 7 days to Oct 1, 2025. The leader here is Claude Code with 5.8M, so Amp scores 14k ÷ 5.8M × 100 = 0. Source
    • VS Code: 038,707 installs of the sourcegraph.amp extension on Sep 15, 2025. The leader here is GitHub Copilot with 40M, so Amp scores 39k ÷ 40M × 100 = 0. Source

    Not counted: GitHub, because it has no public GitHub repo.

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

  13. 13
    pi
    Mario Zechner · open source
    Usage score 0
    • npm 0 · 0
    • GitHub 0 · 8
    • npm: 00 downloads of @mariozechner/pi-coding-agent in the 7 days to Oct 1, 2025. The leader here is Claude Code with 5.8M, so pi scores 0 ÷ 5.8M × 100 = 0. Source
    • GitHub: 08 stars on github.com/badlogic/pi-mono on Sep 20, 2025. The leader here is Gemini CLI with 78k, so pi scores 8 ÷ 78k × 100 = 0. Source

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

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

Not ranked that day, only one source has them: Aider (GitHub, 37k), Droid (npm, 10).

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 Oct 1, 2025

    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

  • GitHubStars · data from Oct 2, 2025

    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 Oct 1, 2025

    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.