TypeSafe workflow evals dashboard

The public dashboard behind the launch numbers: four workflows (security incidents, agent-trace observability, invoice processing, customer service) and ten model configurations compared on accuracy, cost and latency.

Notes

TypeSafe-run workflow evals: the 193.6x / 444.6x multiples and 67.8% overall accuracy; invoice workflow 61.8%.

official — Published by TypeSafe or the project owner. About this label

Destinations

Filed under

• #official
• #benchmark
• #evals

Also on Jevland by TypeSafe

• Introducing System One Models and Jev official · official
• Jev documentation official · official
• The Bitterest Lesson official · official
• typesafe-sdk (Python) repo · official
• @typesafe-ai/sdk (TypeScript) repo · official
• System One adapter for LLM APIs repo · official

In collections

• Start here: official and explainers →
• Benchmarks & independent tests →