Experiments
This is where we will publish measured engineering evidence: same task, different models or architectures; cost and reliability differences; what the failures actually were.
We will not invent results to fill the page. The first FDE experiment write-ups will appear here when there is methodology and data to show.
What a write-up includes
- Architecture under test
- Methodology and evaluation set
- Model and provider
- Token usage, latency, and inference cost
- Success rate and failure modes
- Conclusions that follow from the measurements
- Source or demo links where we can publish them
Published experiments
Coming soon.
Related Labs receipts
These are existing public artifacts from the product and foundations work. They are not customer FDE case studies.