How We Score AI Tools

Last updated: July 29, 2026

Most AI tool comparison sites rank by whoever pays the most. We don't. Our scoring is fully transparent — no black boxes, no paid placements influencing rankings. Here's exactly how it works.

The Formula

Score = (Capability × 0.4) + (Quality × 0.3) + (Price × 0.2) + (Breadth × 0.1)

Each tool gets a score from 0 to 100. Higher is better. The weights reflect what matters most when choosing an AI tool — does it do the job, how well, at what cost, and how versatile is it.

Score Breakdown

Capability Match (40%)

🔍

The heaviest weight. We check if the tool actually supports the features relevant to your chosen use-case (General, Coding, Images, Video, Voice, Research). Only applicable features are counted — an image tool isn't penalized for lacking code execution.

N/A vs Fail: Features not relevant to a tool's category are marked N/A (grey), not counted as a fail. This prevents image generators from being unfairly penalized for lacking coding features.

Quality Rating (30%)

Based on real public benchmarks and community consensus:

  • LLMs: LMArena ELO, MMLU, SWE-bench, HumanEval, GSM8K scores from public leaderboards
  • Image tools: Aesthetic ELO from Hugging Face leaderboards
  • Other tools: Aggregated user ratings and expert reviews

Rating is mapped to 0-100 scale (5.0★ = 100, 4.0★ = 80, etc.)

Price Value (20%)

💰

Cheaper doesn't automatically mean better. We score value-for-money:

  • Free = 100 (best value)
  • Freemium = 85 (has free tier)
  • $ = 70 (under $15/mo)
  • $$ = 55 ($15-50/mo)
  • $$$ = 40 ($50+/mo)

LLM API pricing is auto-updated daily from OpenRouter API.

Breadth (10%)

🎛️

A small bonus for tools that cover more capabilities (image gen + voice + coding = more versatile). This is intentionally the lowest weight — a specialist tool that's great at one thing shouldn't lose to a mediocre generalist.

Category Winners

We don't crown a single "overall winner." Instead, tools can win multiple categories:

  • 🏆 Best for General Chat
  • 🏆 Best for Coding
  • 🏆 Best for Value / Price
  • 🏆 Best for Quality
  • 🏆 Most Versatile

A tool that's amazing at one thing (like Midjourney for images) can win its category even if it's not a great all-rounder.

Data Sources

  • LLM API Pricing: OpenRouter API (auto-updated daily via GitHub Actions)
  • Benchmark Scores: Public leaderboards (LMArena, OpenAI, Anthropic, Hugging Face)
  • Tool Features & Specs: Manually curated from official vendor websites
  • Non-LLM Pricing: Manually verified from vendor pricing pages

Affiliate Disclosure

Some links on our site may be affiliate links. This means we may earn a commission if you click through and make a purchase — at no additional cost to you. Affiliate relationships do not influence our benchmark scores or rankings. The scoring formula is applied identically to all tools regardless of affiliate status.

Questions about our methodology?

Email us at support@myaipicker.com or check our tools yourself:

Try the Compare Deck →