aigratis
iden
deepseek-v4-flashFree/1M×6
nemotron-3-super-120b-a12bFree/1M×3
step-3-7-flashFree/1M×3
nemotron-3-nano-omni-30b-a3b-reasoningFree/1M×3
deepseek-v4-proFree/1M×3
deepseekFree/1M×3
laguna-s-2-1Free/1M×3
gemini-3-6-flashFree/1M×3
laguna-xs-2-1Free/1M×3
hy3-freeFree/1M×3
glm-5-3-flashFree/1M×2
gemma-4-31b-itFree/1M×2
mimoFree/1M×2
lfm-2-5-2-6bFree/1M×2
qwen3-8-27bFree/1M×2
nemotron-3-ultra-550b-a55bFree/1M×2
codestralFree/1M×2
deepseek-v4-flash-visionFree/1M×2
ling-3-0-flash-finFree/1M×2
ling-3-0-flash-santeFree/1M×2
nemotron-3-5-lightningFree/1M×2
gemini-3-5-flash-liteFree/1M×2
hy3Free/1M×2
north-mini-codeFree/1M×2
deepseek-v4-flashFree/1M×6
nemotron-3-super-120b-a12bFree/1M×3
step-3-7-flashFree/1M×3
nemotron-3-nano-omni-30b-a3b-reasoningFree/1M×3
deepseek-v4-proFree/1M×3
deepseekFree/1M×3
laguna-s-2-1Free/1M×3
gemini-3-6-flashFree/1M×3
laguna-xs-2-1Free/1M×3
hy3-freeFree/1M×3
glm-5-3-flashFree/1M×2
gemma-4-31b-itFree/1M×2
mimoFree/1M×2
lfm-2-5-2-6bFree/1M×2
qwen3-8-27bFree/1M×2
nemotron-3-ultra-550b-a55bFree/1M×2
codestralFree/1M×2
deepseek-v4-flash-visionFree/1M×2
ling-3-0-flash-finFree/1M×2
ling-3-0-flash-santeFree/1M×2
nemotron-3-5-lightningFree/1M×2
gemini-3-5-flash-liteFree/1M×2
hy3Free/1M×2
north-mini-codeFree/1M×2
AI handouts · refreshed every 6 hours

Find the AI models you can use for free.

We collect the models AI gateways give away, along with where you can actually use them. Everything below costs nothing. Click any row to see which gateways serve it.

Need a paid model? Turn off the Free filter below, or open the full price board, which lines up every gateway in USD per 1 million tokens.

298
free models
1.6K
models total
24
gateways

1–25 of 298 models

Gateways we track

Every gateway's prices are pulled automatically from its official API, never retyped by hand.

Detail per gateway
See 12 more gateways

Prices here are re-pulled every 6 hours and can change at any time without notice. Before you pay, check the gateway's own pricing page directly.

How are these prices collected?

Every 6 hours we call each gateway's model-list endpoint with our own API key.

Gateways that do not expose prices through an API are filled in by hand: their official pricing page is read by an AI, and the result is checked before it is saved. Numbers from that route are the most likely to be off or out of date, because they only change when their page is read again.

The units differ — some quote per token, some per thousand, some per million, in USD or in Rupiah. Everything is converted to USD per 1 million tokens so it is comparable. Rupiah figures use the daily market rate, not the rate a gateway uses when it bills you.

Model names are normalised too: anthropic/claude-opus-5 and claude-opus-5 are the same model, so they share one row.

More in the Terms of Service, under Price accuracy.