100+ models tracked · 25+ free models · Updated hourly

Stop Overpaying for AI.
Find the cheapest model that actually works.

Compare pricing, speed, and benchmarks across 100+ AI models. Discover free alternatives, calculate your savings, and make data-driven decisions — all in one place. No signup required.

Cheap coding Best Chinese Best Vision Long context Free models Fast translation

Recommended for You

Models tracked
Across all providers
Free models
No payment required
Providers
OpenAI, Google, Anthropic...
90%
Avg. savings
Switch to free/cheap alternatives
Free resources

Free AI models, all in one place

The internet's most complete directory of free LLMs, vision models, and embeddings. No credit card, no signup — just use them.

Browse all free models →
Cost intelligence

Why teams choose ModelHub

Go from "what should we use?" to a shareable, defendable choice — while cutting costs by up to 90%.

Live Price snapshot — per 1M tokens

Loading...
Updated: Loading... Green = cheapest 10% · Red = most expensive 10%
Hands-on testing

Best models by use case

Based on real-world testing across 30+ models — not just benchmarks

Save money

The cost intelligence advantage

See how much you could save by switching to cheaper or free alternatives

Switch from GPT-4o to DeepSeek V3
90%
Cost reduction with similar quality
GPT-4o
$2.50 / $10.00
per 1M tokens
DeepSeek V3
$0.27 / $1.10
per 1M tokens
At 10M tokens/month, you'd save $97,200/year
Switch from Claude to free tier
100%
Cost reduction using free alternatives
Claude Sonnet 4
$3.00 / $15.00
per 1M tokens
Gemini Flash (free)
$0.00
15 RPM free tier
For development & testing, free tiers are often enough. Switch to paid only at production scale.
Coverage

Every major AI provider, tracked

We monitor pricing and capabilities across all leading AI providers — so you don't have to

OpenAI
GPT-4o, o1, o3...
Anthropic
Claude 4, Opus
Google
Gemini, Gemma
DeepSeek
V3, R1
Meta
Llama 3.3
Mistral
Large, Mixtral
Alibaba
Qwen 2.5
Cohere
Command R
Groq
Fast inference
Microsoft
Phi-3
xAI
Grok
Hugging Face
100+ open
Buyer resources

Reference materials you can actually use

Not a company portal. Real checklists, glossaries, and exportable templates for AI model selection.

Procurement pack
Download ready-to-share templates for reviews, approvals, and vendor calls.
Exports as Markdown so teams can edit in Notion, Google Docs, or Git.
Benchmark glossary
Quick definitions for the metrics you'll see across Models, Compare, and Rankings.
What do these benchmarks measure?
  • HumanEval: code generation correctness (unit-test style).
  • MMLU: broad knowledge across many subjects (multiple choice).
  • GPQA: difficult reasoning-heavy QA (often harder than MMLU).
  • GSM8K: grade-school math reasoning.
  • MMMU: multimodal vision + reasoning.
Pricing: why input and output both matter
Output tokens are frequently 3–10× more expensive. Monthly cost must include both sides or "cheap" picks become expensive in production.
Decision workflow
A clean, repeatable path from requirements → shortlist → test → sign-off.
  • Build shortlist in Models (filters + export).
  • Compare finalists in Compare (share link + brief).
  • Validate prompts in Playground (real workloads).
  • Budget check in Pricing before production.

Ready to stop overpaying for AI?

Test any model for free in our playground. No API key, no signup — 10 free requests per hour.