Embedded Arena MCU QA (Direct Prompting): leaderboard
A knowledge test from Embedded Arena (2026): 30 expert-written questions on deploying deep learning models to microcontrollers, covering memory layouts, hardware-specific quantization limits, SDK quirks and troubleshooting, answered from the prompt alone. Accuracy in percent; higher is better.
Metric: Accuracy (%) on the 30 expert-written question-answer pairs about deploying deep learning on microcontrollers (model compression, hardware constraints, framework-specific code and deployment troubleshooting), answered zero-shot in a single turn from the prompt alone; higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 6 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5 | 60 | #167 |
| 2 | Gemini 2.5 Pro | 45 | #266 |
| 3 | GPT-5 Mini | 38.3 | #348 |
| 4 | Gemini 2.5 Flash | 33.3 | #453 |
| 5 | GPT-5 Nano | 31.7 | #889 |
| 6 | Gemini 2.5 Flash Lite | 16.7 | #958 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=embedded-arena-mcu-qa-direct-prompting · How It Works · Data refreshed daily, snapshot 2026-11-13.