firasd/vibesbench
Conversational AI benchmark
57 best points · 104 comments · 6 Hacker News posts · 12 GitHub stars · View on GitHub
Hacker News discussions
- AI sycophancy panic (57 points, 102 comments, by firasd, 2026-01-04)
- Vibesbench: Prompts to track conversational regression in AI models (3 points, 0 comments, by firasd, 2026-01-01)
- Anthropic silently rewriting Claude punctuation output in API (2 points, 1 comments, by firasd, 2026-01-07)
- Claude output silently rewritten by Anthropic (2 points, 0 comments, by firasd, 2026-01-09)
- Vibesbench: A Multi-Turn Conversational AI Benchmark (1 points, 0 comments, by firasd, 2025-12-25)
- Stockfish shows Morphy and Fischer didn't sacrifice their queens (1 points, 1 comments, by firasd, 2025-12-27)