Skip to content

Evaluation

Request Access

Evaluate ISIRO Runtime on your workload without sharing your model.

Evaluation paths

Two ways to evaluate

We currently support BF16 vLLM workloads on NVIDIA GPUs in on-prem and cloud environments. Support for additional inference frameworks and hardware platforms is on the roadmap.

Developer evaluation

Technical fit on a representative workload, including memory-traffic reduction, output equivalence, and framework and hardware compatibility.

Pilot evaluation

A more detailed, structured benchmark in your environment with integration support, before production decisions.

What happens next

  • We promptly review your request.

  • We follow up with next steps for your evaluation path.

Request access

Evaluation path *

Your details

Workload details

Primary interests *

Prefer email? hello@isiro.ai