evaly demo

Browse real benchmark results without any setup or API keys.

evaly demo [CASE]

The demo command opens curated benchmark reports in your browser. It requires no API keys, no configuration, and no generation — just real data from 9 models scored by Gemini 2.5 Flash.

Zero friction: This is the recommended first command after pip install evalytic. See what Evalytic can do before setting up API keys.

Usage

# Open the full showcase page (4 case studies)
evaly demo

# Open a specific case study
evaly demo flagship
evaly demo face
evaly demo prompt-trap
evaly demo product

Case Studies

CaseTitleModelsKey Finding
flagship The $0.003 Question Flux Schnell, Dev, Pro, Recraft, Ideogram Same price band, 0.5-point quality gap
face That's Not Me 5 img2img models + ArcFace metric r=0.99 VLM↔metric correlation
prompt-trap The Prompt Trap Simple vs detailed prompts Visual quality tied, fidelity ≠ tied
product Is That Still My Product? Seedream, Flux Kontext, FireRed, Reve Input fidelity as the differentiator

All benchmarks use real images generated via fal.ai and scored by Gemini 2.5 Flash. Total cost: $0.55.

What You See

Running evaly demo does two things:

  1. Terminal teaser — a Rich-formatted table showing all 4 case studies with key findings
  2. Browser — opens the interactive showcase page (or a specific report) with images, scores, and analysis

Next Steps

After exploring the demos, run your own benchmark:

# Setup wizard (collects API keys)
evaly init

# Your first benchmark
evaly bench -m flux-schnell -p "A cat in a top hat" -y