Celeris-1 Decision Launches, Tops Benchmarks

This is a practical infrastructure release, not another chatbot—the value is turning model judgments directly into routable application logic.

• –System One API compatibility lets Jev users migrate by changing the endpoint, key, and model name.
• –Typed probabilities can power classification, routing, filtering, scoring, and verification without parsing prose.
• –Image inputs and optional explanations broaden use cases to receipts, documents, screenshots, and visual review workflows.
• –The benchmark claims are promising, but independent methodology and full results are needed to assess generalization and calibration.
• –Explanations add latency, so production systems should reserve them for human-review paths. ([Celeris Decision](https://celeris.ai/decisions), [Celeris documentation](https://docs.celeris.ai/decisions))