Benchmark

Run a prompt across providers simultaneously — compare TTFB, throughput, and cost.

Prompt preset

Runs per provider — 3

110

Providers

Groq

Llama 3.3 70B

OpenAI

GPT-4o mini

Anthropic

Claude Haiku

Local Demo

Project Assistant

Ollama

Llama 3.2

NVIDIA NIM

Nemotron 70B

Results

Configure providers and click ⚡ Run Benchmark to start.

How it works: Requests are proxied server-side (no CORS issues). RAIS, OpenAI, Anthropic, and Groq response formats are all detected automatically for token counting. API keys are held in-memory only and never stored.