license	other
pretty_name	Razavi Bench
language	en
task_categories	question-answering visual-question-answering
tags	analog-design circuit-design benchmark multimodal llm-evaluation electronic-design-automation
size_categories	n<1K
configs	config_name data_files tasks split path train data/tasks.jsonl

Razavi-Bench

Visit our website

Razavi-Bench is an expert-curated benchmark for analog-design reasoning: 50 multimodal questions from Behzad Razavi's Analog Design Experiments With AI (Part 1: 30 questions, Part 2: 20), each with a figure where needed and a curated golden answer under tasks/. The leaderboard, every model answer, and the judge's scores are on the website.

To evaluate a model yourself, see tools/README.md: one OpenRouter key runs the 50 tasks, grades them with the DeepSeek V4 Pro judge, and builds the website data.

Citation

@misc{zhang2026razavibench,
  title        = {Razavi-Bench: An Expert-Curated Benchmark for Analog-Design Reasoning},
  author       = {Zhishuai Zhang},
  year         = {2026},
  howpublished = {\url{https://github.com/Arcadia-1/razavi-bench}},
  url          = {https://razavi-bench.tokenzhang.com/},
  note         = {Benchmark repository}
}

License

Software code is licensed under Apache-2.0. Benchmark materials (questions, figures, golden solutions, rubrics, model outputs, and scores) adapt Behzad Razavi's articles with his permission and are available for viewing, citation, non-commercial research, and local evaluation only; they may not be redistributed, used for training, or incorporated into other benchmarks or datasets without prior written permission. See LICENSE for the full terms.

References

• B. Razavi, "Analog Design Experiments With AI—Part 1 [The Analog Mind]," in IEEE Solid-State Circuits Magazine, vol. 17, no. 4, pp. 11-15, Fall 2025.
• B. Razavi, "Analog Design Experiments With AI—Part 2 [The Analog Mind]," in IEEE Solid-State Circuits Magazine, vol. 18, no. 2, pp. 8-13, Spring 2026.