Interactive Playground
Benchmark Kervalt models in a web sandbox.
Compare Tafari Core, Zuberi Vector, and Sankofa Rank side by side — tune parameters, inspect outputs, and validate quality before you write a line of integration code.
Prompt
Output
412 tokens · 1.8s · eu-centralThe agreement grants the vendor broad rights to process customer data on infrastructure outside the EEA, including sub-processors in non-equivalent jurisdictions. Two clauses — 7.2 (cross-border transfer) and 9.4 (model training on customer inputs) — conflict with EU sovereignty requirements and should be renegotiated before signature. All other obligations fall within standard commercial terms.
Parameters
This is a static preview of the playground. Live inference requires an API key.
Evaluate before you integrate
One sandbox for the whole Kervalt stack
Tafari Core
Test reasoning, summarization, and tool-use quality on your own prompts. Measure latency and token throughput from EU regions.
Zuberi Vector
Embed sample documents and inspect vector output dimensions, similarity scores, and multilingual retrieval behavior.
Sankofa Rank
Paste candidate passages and watch reranking reorder them by relevance — the fastest way to validate retrieval quality gains.
Ready for real inference?
The playground previews a fraction of what the API exposes. Get an API key and run the same models against your own data — in your cloud or ours.
Ready to deploy sovereign AI?
Run Kervalt models in your own cloud, on your own terms.