Surprising screenshot - Most token usage is non-coders (openrouter ranking)
Surprising ranking reveals 6 of top 10 'coding' agents are actually non-coding applications.
A viral screenshot from OpenRouter's public analytics dashboard has revealed a counterintuitive trend in how developers are using AI models. Despite the category label of 'coding agents,' six of the top ten applications by token usage are actually non-coding tools being repurposed for programming tasks. Specialized coding models like OpenCode don't even appear in the top rankings, suggesting that general-purpose models are winning the practical adoption battle.
The data indicates that versatile models like Hermes and OpenClaw are being heavily used for coding work, potentially because they offer broader capabilities beyond just code generation. This trend challenges the assumption that specialized tools always dominate their niche and suggests developers value flexibility over specialization. The community discussion highlights how users are adapting general AI assistants to handle coding tasks effectively, raising questions about the future of single-purpose AI tools.
- OpenRouter analytics show 6 of top 10 'coding agents' are actually non-coding applications
- Specialized coding model OpenCode doesn't rank in top 10 usage charts
- General-purpose models like Hermes and OpenClaw are being repurposed for coding tasks
Why It Matters
This suggests developers prefer versatile AI tools over specialized ones, potentially shifting how AI coding assistants are built and marketed.