Skip to content
fedi.software

AI models, tools and skills

Leaderboards with confidence intervals, prices per unit with their history, free APIs, apps to download and agent skills — every number with its source and date.

fedi-index: top models

View all

General

  1. 1 S Claude Opus 5.5 1,504
  2. 2 S Claude Fable 5.1 1,498
  3. 3 S Claude Opus 5 1,493
  4. 4 A Gemini 4 Argon (High) 1,490
  5. 5 S Claude Fable 5 1,484

Language: English provisional

  1. 1 A Gemini 4 Argon (High) 1,511
  2. 2 A Claude Opus 4.6 1,488
  3. 3 A Claude Opus 5.5 1,485
  4. 4 A Claude Opus 5 1,480
  5. 5 A Claude Fable 5.1 1,478

AI infographics

View all
Model for a task Free alternatives Will it run locally? Model types Builds for local AI AI models by year Free AI limits AI tools map fedi-index

Top text models

View all

Updated · Sources: LMArena, LiteLLM, models.dev

Tier Model Elo Input / output, per 1M Context App
S1,533— —
S1,512$4 / $20 1M
S1,511$10 / $50 1M
S1,507$5 / $25 1M
S1,503$4.30 / $21 1M

Free APIs

View all
  • S Kimi K3 Nvidia Free
  • S GLM 5.3 Nvidia Free
  • S GLM 5.2 Nvidia Free
  • S GLM 5.3 Flash Nvidia Free
  • S DeepSeek V4.1 Flash Nvidia Free

Trending skills

View all
  • ui-taste uizze.sh 441,477
  • design-mobile-apps designed-by-ai/skills 657,107
  • ai-image-generation magentosh/superpowers 308,602
  • twitter-automation magentosh/superpowers 308,313
  • ai-video-generation magentosh/superpowers 308,455

Tools

API providers OpenRouter · Anthropic API · OpenAI API · Google AI Studio VPS for AI Hetzner · Vultr · Linode (Akamai) · DigitalOcean Local LLM apps Ollama · LM Studio · Jan · GPT4All Automation n8n · Make · Zapier · Activepieces Coding agents Claude Code · OpenAI Codex CLI · Gemini CLI · Cursor AI apps ChatGPT · Claude · Gemini · Microsoft Copilot

What this section covers

The AI section of fedi.software collects the numbers you need when choosing and using AI models, in one place and with their origin shown. Rankings come from a public leaderboard, LMArena, and prices with context windows from models.dev and LiteLLM. We do not run benchmarks ourselves; we aggregate, add our own tiers and keep the history of prices.

The overview page shows only a preview of each part: the top 5 text models by tier and rating, five free APIs, five trending agent skills and the tool categories. The full tables, with filters, a cost calculator and price trends, are one click away.

Where to go next

How to read the numbers

Every table carries a line above it with the date of the data and its sources. Ratings and API prices are refreshed daily, subscription prices of apps monthly, VPS plans and GitHub stars weekly. The tier letters S to D are ours, not the sources': they describe where a model stands within one leaderboard once statistical uncertainty is taken into account, which is why two models can share a tier even when their ratings differ by a few points.

Keep in mind that a leaderboard reflects blind votes of arena users on the prompts they happen to send. It says which answers people preferred, not how a model will handle your particular task, and the order at the top changes often as new models arrive. Treat a rank as a snapshot of one day and check the date next to it.

Partner links

Some tool cards link to a vendor through a partner programme. Such links are marked "Partner link" and pass through a redirect on our domain. They never affect tiers, the order of models or which tools we list.