The Benchmark
Measuring what matters in firmware.
The V1 benchmark is a work in progress. We're consulting with firmware engineers across a variety of domains to develop a benchmark which reflects the entire industry.
Tasks
117
+23 this month
Domains
8
expanding
Avg. Pass Rate
31%
across all models
Contributors
12
engineers & researchers
Model Accuracy
Fieldwork Benchmark Results
Live — updated daily
Model
0%15%30%45%60%
Claude 4 Opus
47%
GPT-4.1
42%
Gemini 2.5 Pro
38%
Claude 4 Sonnet
35%
GPT-4.1 mini
28%
Gemini 2.5 Flash
24%
Claude 4 Haiku
19%
GPT-4.1 nano
12%
Pass rate on Fieldwork V1 benchmark tasks. Higher is better. Scores are mockup data — final results will be published with the V1 release.
Coverage
Industry Domains
Robotics
24
tasks
Audio / DSP
18
tasks
Medical Devices
15
tasks
Aerospace
12
tasks
IoT / Smart Home
21
tasks
Manufacturing
9
tasks
Drones / UAV
11
tasks
Farming / Agri
7
tasks
Interested in contributing?
We're consulting with firmware engineers across a variety of domains. We'd love to put your name on our publication.
Email tk@fieldwrk.ai