Free report·
AI Coding Tools Benchmark
How OpenAI Codex, Claude Code, and Cursor compare in terms of adoption, engagement, satisfaction, and productivity gains among software developers.
OpenAI Codex is the most widely adopted tool, used by nearly 60% of professional developers writing AI-assisted code, and leads on satisfaction with a CSAT score of 88 out of 100.
This preview benchmarks three prominent AI coding tools — OpenAI Codex, Claude Code, and Cursor — across adoption, usage intensity, developer satisfaction, and productivity gains. Based on a global survey of 800+ professional developers who write code with AI assistants, agents, or AI-native IDEs, it maps how each tool performs on measurable outcomes rather than crowning a single winner. Codex emerges as the broad platform play, with the highest adoption, usage intensity, and satisfaction. Claude Code, though less widely adopted, edges ahead on pull request throughput and weekly time savings. Cursor shows heavy engagement among its users but weaker productivity outcomes, indicating room for optimisation. The report also combines PR throughput and time savings into a single return-on-investment measure, classifying developers into elite, high, medium, and low productivity bands. It is a lighter, freely available version of a more comprehensive benchmark covering the full range of AI coding tools, due in December 2025.
Key findings
- Codex is the most widely used tool, with nearly 60% of AI-assisted developers relying on it, followed by Claude Code at 38% and Cursor at 24%.
- On usage intensity, Codex leads with 57% of users relying on it for at least half their coding tasks, ahead of Cursor at 52% and Claude Code at 48%.
- Codex scores 88 on overall CSAT (0-100 scale), while Claude Code and Cursor both score 78.
- Claude Code leads on weekly time savings, with 67% of users saving at least six hours per week, ahead of Codex at 65% and Cursor at 59%.
- 79% of Claude Code users report increased merged PR volume, ahead of Codex at 77% and Cursor at 68%; Codex and Claude Code are tied at 45% of users in the high or elite productivity group, versus a market median of 31%.
Questions this report answers
Which AI coding tool has the highest adoption among developers?
OpenAI Codex is the most widely adopted, used by nearly 60% of professional developers writing AI-assisted code, compared with 38% for Claude Code and 24% for Cursor.
Which AI coding tool delivers the best productivity gains?
Claude Code edges ahead on outcome metrics: 67% of its users save at least six hours per week and 79% report higher merged PR volume, though it is tied with Codex at 45% of users in the high or elite productivity group.
How satisfied are developers with Codex, Claude Code, and Cursor?
Codex scores highest on overall CSAT at 88 out of 100, while Claude Code and Cursor both score 78, with Codex outperforming on writing new code, modifying existing code, and debugging.
Does heavy use of an AI coding tool guarantee higher productivity?
No. Cursor commands relatively intense use (52% of users use it for at least half their work) yet yields a smaller high-productivity segment (28%) below the market median of 31%, showing productivity does not scale linearly with usage intensity.
How does SlashData define productivity for AI coding tools?
Developers are classified into elite, high, medium, and low bands based on both pull request throughput and weekly time savings, combined into a single scored measure ranging from 2 to 8 points.
Who it's for
- AI technology vendors aligning product strategy with measurable developer outcomes
- Engineering leaders and technology buyers selecting or standardising AI coding tools
- Developer relations and product teams benchmarking tool adoption and engagement