Ciited

ReportsAI Coding Tools

Industry report

AI Coding Tools

Which AI coding assistants and agents do AI models recommend? Recommendation rates from a fixed set of buyer prompts across ten models.

Last run Nov 15, 20262,500responses10 models

Rankings describe what AI models answered to a fixed set of prompts. They are not endorsements or quality ratings. Product names belong to their owners.

Latest run · Nov 15, 2026

Recommendation leaderboard

How often each product was recommended in the latest run.

Open this runAll runs

  • Rank 01= · Leader

    GitHub Copilot

    78.9%

    71–87%

    ——
  • Rank 02=

    Cursor

    66.3%

    58–74%

    ——
  • Rank 03

    Claude Code

    46.7%

    40–53%

    ——
  • Rank 04=

    Windsurf

    33.7%

    28–39%

    ——
  • Rank 05=

    Amazon Q Developer

    30.1%

    24–37%

    ——
  • Rank 06=

    Aider

    27.1%

    19–35%

    ——
  • Rank 07=

    OpenAI Codex

    24.3%

    20–29%

    ——
  • Rank 08=

    Continue

    22.8%

    15–30%

    ——
  • Rank 09=

    Cline

    22.1%

    15–30%

    ——
  • Rank 10=

    JetBrains AI Assistant

    20.3%

    14–27%

    ——
  • Rank 11=

    Tabnine

    19.5%

    14–25%

    ——
  • Rank 12=

    Gemini Code Assist

    16.0%

    11–21%

    ——
  • Rank 13=

    Sourcegraph Cody

    15.7%

    11–21%

    ——
  • Rank 14=

    Devin

    13.6%

    10–17%

    ——
  • Rank 15=

    Roo Code

    12.5%

    7–18%

    ——
  • Rank 16=

    OpenHands

    11.4%

    8–15%

    ——
  • Rank 17=

    Google Jules

    8.7%

    5–12%

    ——
  • Rank 18=

    Qodo

    7.7%

    4–12%

    ——
  • Rank 19=

    GitLab Duo

    7.6%

    3–13%

    ——
  • Rank 20=

    Junie

    7.1%

    4–12%

    ——
  • Rank 21=

    Gemini CLI

    6.2%

    3–11%

    ——
  • Rank 22=

    Tabby

    5.7%

    1–12%

    ——
  • Rank 23=

    SWE-agent

    5.5%

    3–8%

    ——
  • Rank 24=

    Augment Code

    5.0%

    2–8%

    ——
  • Rank 25=

    Goose

    4.7%

    1–10%

    ——
  • Rank 26=

    OpenCode

    3.9%

    1–8%

    ——
  • Rank 27=

    Factory

    3.4%

    2–5%

    ——
  • Rank 28=

    Supermaven

    3.0%

    1–5%

    ——
  • Rank 29=

    Zed

    3.0%

    1–5%

    ——
  • Rank 30=

    Amp

    2.4%

    1–4%

    ——
  • Rank 31=

    Sentry Seer

    2.1%

    0–4%

    ——
  • Rank 32=

    GitHub Copilot Workspace

    1.6%

    1–3%

    ——
  • Rank 33=

    Kilo Code

    1.6%

    0–3%

    ——
  • Rank 34=

    Sweep

    1.2%

    0–2%

    ——
  • Rank 35=

    IBM watsonx Code Assistant

    1.1%

    0–3%

    ——
  • Rank 36=

    Kiro

    1.1%

    0–2%

    ——
  • Rank 37=

    Qodo Gen

    1.1%

    0–2%

    ——
  • Rank 38=

    Rovo Dev

    1.1%

    0–2%

    ——
  • Rank 39=

    Plandex

    1.0%

    0–3%

    ——
  • Rank 40=

    Tembo

    1.0%

    0–2%

    ——
  • Rank 41=

    Void

    1.0%

    0–3%

    ——
  • Rank 42=

    mini-SWE-agent

    0.9%

    0–2%

    ——
  • Rank 43=

    Twinny

    0.9%

    0–2%

    ——
  • Rank 44=

    Crush

    0.8%

    0–2%

    ——
  • Rank 45=

    Qwen Code

    0.8%

    0–2%

    ——
  • Rank 46=

    Codegen

    0.7%

    0–2%

    ——
  • Rank 47=

    Warp

    0.7%

    0–2%

    ——
  • Rank 48=

    Refact.ai

    0.7%

    0–2%

    ——
  • Insufficient evidence

    Open Interpreter

    0.5%

    0–2%

    ——
  • Insufficient evidence

    Cosine Genie

    0.5%

    0–1%

    ——
  • Insufficient evidence

    AutoCodeRover

    0.4%

    0–1%

    ——
  • Insufficient evidence

    CodeGPT

    0.4%

    0–1%

    ——
  • Insufficient evidence

    Murmur

    0.3%

    0–1%

    ——

Recommendation rate counts a product once per response. All-models figures are the unweighted mean of per-model rates. A product is ranked only when it was recommended in at least 10 responses; otherwise the rank is insufficient evidence. Products share a rank when their 95% intervals overlap with the product above.

How the recommendation rate is moving

Top products by recommendation rate · one point per published run

A trend appears once two runs have been published for this slice.

Each point is one published run. The line is recommendation rate, not share of all recommendations.

A different leaderboard in every model

Top 5 products by recommendation rate · all topics

Same prompts. Separate scores.

Claude Sonnet 5.5

n = 250

No web search

  1. GitHub Copilot76%
  2. Claude Code73%
  3. Cursor71%
  4. Windsurf35%
  5. Amazon Q Developer32%

DeepSeek V4.1 Flash

n = 250

No web search

  1. GitHub Copilot84%
  2. Cursor60%
  3. Claude Code39%
  4. Continue38%
  5. Windsurf38%

Gemini 4 Hydrogen

n = 250

No web search

  1. GitHub Copilot86%
  2. Cursor77%
  3. Windsurf65%
  4. Claude Code56%
  5. Aider40%

Muse Spark 1.3

n = 250

No web search

  1. GitHub Copilot84%
  2. Cursor61%
  3. Amazon Q Developer53%
  4. Windsurf49%
  5. Tabnine36%

Mistral Large 4

n = 250

No web search

  1. GitHub Copilot75%
  2. Cursor61%
  3. Amazon Q Developer42%
  4. Windsurf39%
  5. Aider34%

Kimi K3.1 Pro

n = 250

No web search

  1. GitHub Copilot76%
  2. Cursor56%
  3. Claude Code34%
  4. Continue28%
  5. Windsurf24%

GPT-6.1 Sol

n = 250

No web search

  1. GitHub Copilot80%
  2. Cursor64%
  3. Claude Code42%
  4. OpenAI Codex31%
  5. Continue20%

Perplexity Sonar

n = 250

Web search on

  1. GitHub Copilot65%
  2. Claude Code54%
  3. Cursor50%
  4. OpenAI Codex22%
  5. Aider21%

Grok 4.7

n = 250

No web search

  1. GitHub Copilot83%
  2. Cursor74%
  3. Claude Code36%
  4. Amazon Q Developer34%
  5. Windsurf31%

GLM-5.3

n = 250

No web search

  1. GitHub Copilot72%
  2. Cursor61%
  3. Claude Code42%
  4. Windsurf38%
  5. Aider33%

Also recommended across AI coding tools

Products from other categories that came up in the same answers. The prompts ask about coding assistants and agents, so these rates show incidental recommendations, not a full ranking of each category.

AI app builders

  1. Replit Agent11%
  2. v08%
  3. Lovable8%
  4. Bolt.new8%
  5. FlutterFlow2%

AI code review

  1. CodeRabbit8%
  2. Greptile6%
  3. Qodo Merge4%
  4. Graphite3%
  5. Sourcery2%

Code security

  1. Snyk5%
  2. Semgrep4%
  3. SonarQube4%
  4. GitHub Advanced Security3%
  5. Checkmarx One2%

Testing & QA

  1. Diffblue Cover4%
  2. mabl2%
  3. Momentic2%
  4. Qodo Cover2%
  5. testRigor2%

Migration & refactoring

  1. Moderne2%
  2. AWS Transform2%
  3. Codemod1%
  4. Grit1%
  5. IBM watsonx Code Assistant for Z1%

Biggest movers this run

No earlier published run to compare.

Movers appear after a second run is published.

Cited domains

Domains linked in model responses for this run.

  • github.com

    109citations

  • dev.to

    80citations

  • superblocks.com

    75citations

  • zapier.com

    56citations

  • kilo.ai

    55citations

  • learn.g2.com

    55citations

  • aibusinessweekly.net

    51citations

  • ibm.com

    43citations

Built to be repeated

Methodology v1
Fixed prompt set
50 prompts

Version 1. Editing a prompt starts a new series.

Topics
5 topics

Equal weight. A topic with fewer answers does not count less.

Sampling
5 samples / prompt / model

2,500 responses in this run.

Models
10 models, equal weight

Claude Sonnet 5.5, DeepSeek V4.1 Flash, Gemini 4 Hydrogen, Muse Spark 1.3, Mistral Large 4, Kimi K3.1 Pro, GPT-6.1 Sol, Perplexity Sonar, Grok 4.7, GLM-5.3

Minimum evidence
10 recommendations

Below that, the rank is insufficient evidence.

Recommendation rate is the share of valid responses that recommend a product, counted once per response. The all-models figure is the unweighted mean of the per-model rates, so a longer answer does not get a heavier vote. 95% intervals come from a bootstrap that resamples prompts, not individual answers. These answers come from provider APIs with pinned model slugs. They are not the consumer apps, which add their own system prompts, memory, and search. Lowest valid-response rate in this run: 100%.

Sample prompt · Autonomous agents
“Which AI coding agents can take a GitHub issue and open a working pull request on their own?”
Sample prompt · Debugging & testing
“What AI tools are best for debugging failing tests and fixing bugs in a Python project?”

Free audit

Where does your brand rank?

These reports cover a category. A free audit grades your own site against the same four models.

Run a free audit