Best free tier for inference

LPU-hosted Llama + Mixtral with a free rate-limited API Console tier; very low latency.

Bandwidth

n/a

Requests

Free API with rate limits (e.g. 30 req/min, daily token cap)

Paid from

—

Serverless OpenAI-compatible endpoints for 200+ open models; $5 signup credit, then pay-per-token.

Bandwidth

n/a

Requests

$5 free credits on signup (serverless endpoints)

Paid from

—

Serverless host for open + fine-tuned models with OpenAI-compatible API; $1 signup credit, then pay-per-token.

Bandwidth

n/a

Requests

$1 free credit on signup (serverless endpoints)

Paid from

—

Run thousands of community + open models via a single API; trial credit, then pay-per-second GPU billing.

Bandwidth

n/a

Requests

Free trial credit for new accounts

Paid from

—