Documentation Index
Fetch the complete documentation index at: https://docs.neurometric.ai/llms.txt
Use this file to discover all available pages before exploring further.
Inference Studio
Neurometric Inference Studio is an internal tool to evaluate AI model cost vs performance. It helps you understand what you’re spending and where Small Language Models (SLMs) could replace expensive frontier models.Key Features
- Model Comparison: Run the same prompts across multiple models and compare results side-by-side.
- Quality Scoring: Automated pass/fail quality scoring and cosine similarity comparisons across outputs.
- Cost-Quality Analysis: Interactive scatter plots to visualize the cost-quality tradeoff for each model.
- Background Jobs: Asynchronous processing using Graphile Worker for analyzing large datasets and running Ray jobs.
- Langfuse Integration: Sync traces from your existing Langfuse observability setup directly into the studio.
Tech Stack
The Inference Studio is built with a modern stack:- Next.js 16 (App Router)
- React 19
- Tailwind CSS v4
- D3.js (for scatter plots)
- Prisma (ORM)
- PostgreSQL
- Graphile Worker (background jobs)