Databricks opens its cost-vs-quality Pareto benchmark powering the Unity Gateway Smart Router
pwendell · x · 2026-11-17
Databricks VP of engineering Phil Wendell published the internal benchmark used to measure the cost-vs-quality Pareto frontier for technical work (Databricks AI Index, Nov 2026 update): cost per task on a log scale ($0.05–$10+) vs pass rate, showing where models sit on the frontier.
It powers the Unity Gateway Smart Router, which picks task-appropriate models and will be updated regularly. Context from the same thread: open-source models grew from near zero to 30% of enterprise token share in five months, and P99 weekly spend is $700 vs a fast-growing $2.50 median.
More from Infra
- Evolving Infra beats NVIDIA Dynamo on agent workloads: 4.1x lower latency, 14x lower P95 TTFT — YouJiacheng · 2026-11-17
- Open-source models hit 30% enterprise token share in 5 months, up from near zero — pwendell · 2026-11-17
- Datology claims 6x compute multiplier from data curation alone; 12B beats 30B baseline — pratyushmaini · 2026-11-17
- Together AI's Five-Step Playbook for Migrating From Closed to Open Models — togethercompute · 2026-11-17
- Why tensor parallelism breaks RNG reproducibility in model init — ezyang · 2026-11-17
- Bittensor's Engy tops Venice's AI inference volume despite 28x smaller valuation — markjeffrey · 2026-11-17