SAT baseline
Tiny SAT and UNSAT smoke instances completed in a little over `4.1 s` end to end from upload to final result on the remote SAT worker.
Evidence
Benchmarks and technical case studies
This page collects concrete performance notes, sample workloads, and practical guides for Both Hemispheres. The goal is to give technical buyers something better than generic marketing copy: measured examples, workflow details, and clear next steps.
Measured Examples
Current benchmark notes
LLM latency baseline
A short OpenAI-compatible Llama 4 Scout chat request completed in `3.25 s` total with a `70-token` response payload.
CNF workflow guide
A practical how-to for uploading DIMACS files, interpreting SAT or UNSAT results, and using sample inputs to evaluate the service quickly.
Method
What these pages are and what they are not
- These are real public-service measurements taken against `probqa.com`.
- They are operational baselines, not maximum-throughput claims.
- They are meant to reduce buyer uncertainty and show the service is live.
- For custom workloads, contact support and request a trial benchmark.
Next Step
Use the examples, then run your own test
The fastest path to conversion is still a real workload. Start with trial credits, use a sample file or the curl example, and compare the service against your actual use case.