Zero of twelve models named Windsurf first on the direct prompt; two named Cursor. Windsurf was named by eleven of the twelve models and Cursor by twelve and Windsurf carries 26 labels and Cursor 52, so the shares are not directly comparable.
Named in one category this edition.
Named in one category this edition.
Share is the count of first choices across the direct, paraphrase, budget and scale prompts over all twelve models, for a mid-market B2B company; rank is within the category; every quote names the model and the prompt it came from. Both figures come from the AI coding assistants page.
Bold names in an answer are the products the judge labeled a first choice; a model naming several gives each of them that label. The full answer text for every row is in the record.
Every negative label with a quote, up to three, then the highest-weighted positives, up to three. Six of eight in this category shown.
“avoid standard free-tier or public instances of major AI coding assistants (such as GitHub Copilot standard, Cursor, or Windsurf)” Qwen 3.7 Flash · negative prompt · hard negative
“Listed among tools impacted by IDEsaster; vulnerable to the same chaining of prompt injection with IDE features” GLM 4.7 FlashX · negative prompt · hard negative
“Avoid: Free tiers of Copilot/Windsurf for proprietary code” DeepSeek V4 Flash · negative prompt · hard negative
“Codeium (Best Free Option) ... Codeium's free tier or GitHub Copilot's $10/month plan offer the best balance of cost and functionality.” Claude Haiku 4.5 · budget prompt · first choice
“recommend starting with GitHub Copilot ... or Codeium (for cost efficiency and privacy options) as pilots” Claude Haiku 4.5 · paraphrase prompt · first choice
“Windsurf is widely considered one of the best AI coding assistants for companies with a limited budget” Qwen 3.7 Flash · budget prompt · first choice
Every negative label with a quote, up to three, then the highest-weighted positives, up to three. Six of eight in this category shown.
“avoid standard free-tier or public instances of major AI coding assistants (such as GitHub Copilot standard, Cursor, or Windsurf)” Qwen 3.7 Flash · negative prompt · hard negative
“Over 30 vulnerabilities reported; 100% of tested AI IDEs were vulnerable to chained prompt injection and auto-approved tool calls” GLM 4.7 FlashX · negative prompt · hard negative
“Documented security incidents involving insufficient boundaries and autonomous actions” Kimi K2 · negative prompt · hard negative
“If you want a state-of-the-art, everyday editor that manages your entire workspace effortlessly: Go with Cursor.” Gemini 3.5 Flash · comparative prompt · first choice
“likely to be GitHub Copilot Business or Cursor Business, depending on the specific needs and budget” Llama 4 Maverick · direct prompt · first choice
“the best AI coding assistants are GitHub Copilot Business and Cursor Business” Mistral Small · direct prompt · first choice
Comparisons are drawn for the top three products in each category. The output is the models' output; nothing here is a recommendation by the index.