3,360 experiments show LLMs steer 79% of API picks to big clouds, 70% when coding
santiviquez · x · 2026-11-17
@santiviquez ran 3,360 experiments testing which document processing APIs models recommend: when asked directly, 79% of recommendations went to big-cloud APIs (Azure Document Intelligence, Google Document AI, Gemini API, AWS Textract). When asked to write implementation code, models chose those APIs as the best option 70% of the time.
The data reveals systematic bias in agent-driven tool selection: verbal recommendations and actual coding behavior diverge, with self-preference amplifying once models write code.
More from Models
- Models favor their own APIs in code: Opus calls Claude API in 97% of implementations — santiviquez · 2026-11-17
- Search behavior gap: Opus searched in 10/60 answers, GPT models searched every time — santiviquez · 2026-11-17
- Web search shifts LLM API picks: big clouds fall from 79% to 60% of recommendations — santiviquez · 2026-11-17
- 79% of 2,880 LLM API recommendations went to just four big-cloud services — santiviquez · 2026-11-17
- Hy4 Flash hands-on: DeepSeek-V4.1-level quality at ~1/15 the cost, 60% faster — IlovedamemesUwU · 2026-11-17
- Memory eval leaderboards are misleading: OmniMemEval mixes conditions and tunes on its own benchmark — Snoo-bedooo · 2026-11-17