Chat and agents
Call tools, stream replies and send images to vision models.
POST /openai/v1/chat/completionsLiquid API
Frontier-grade intelligence at a fraction of the compute.
Models for chat, vision, audio and decisions, behind an API that works with your existing code.
curl https://api.liquid.ai/openai/v1/chat/completions \ -H "Authorization: Bearer $LIQUID_API_KEY" \ -H "Content-Type: application/json" \ -d '{ "model": "lfm2.5-24b-a2b", "messages": [{ "role": "user", "content": "Summarize: my order #4821 arrived damaged." }] }'Available on the API
All modelsWhat the API supports
Voice
Take speech in and reply with speech, streamed as it’s generated.
POST /openai/v1/chat/completionsBatch jobs
Run large jobs within 24 hours. Available on request.
POST /openai/v1/batchesDecisions
Get a calibrated probability for every answer, with no text to parse.
POST /decisions/v1/systemoneDecision models
Ask a question. Get a probability.
Decision models read text, images or both and return a calibrated probability for every possible answer.
“I was charged twice for my order. Please refund one of the payments.”
Is this a complaint?
94%YesWhich team should handle it?
71%BillingShipping 22% · Technical 7%
How urgent is it?
3.9out of 5
From $0.02
per million input tokens. Decision models generate no output tokens, so input is all you pay for.
#1
d1-ultra is the top-scoring decision model on two independent leaderboards.
Same SDK
Already calling decision models with the TypeSafe SDK? Point it at api.liquid.ai and keep the rest of your code.