nav.announce

Models

Call several model families with one API key, and connect them to live sensing data.

ModelContextInput / 1M tokensOutput / 1M tokensCapabilityStatus
openai/<flagship>openai/<flagship>———ChatTool callingBeta
Try in chat
deepseek/deepseek-chatdeepseek/deepseek-chat———ChatTool callingBeta
Try in chat
meta-llama/<largest>meta-llama/<largest>———ChatTool callingBeta
Try in chat
openai/<mini>openai/<mini>———ChatTool callingBeta
Try in chat
openai/<reasoning>openai/<reasoning>———ChatReasoningTool callingBeta
Try in chat
deepseek/<reasoner>deepseek/<reasoner>———ChatReasoningTool callingBeta
Try in chat
meta-llama/<small>meta-llama/<small>———ChatBeta
Try in chat
qwen/<max>qwen/<max>———ChatTool callingUnder evaluation
Try in chat
qwen/<flash>qwen/<flash>———ChatUnder evaluation
Try in chat
qwen/<thinking>qwen/<thinking>———ChatReasoningUnder evaluation
Try in chat
anthropic/<claude>anthropic/<claude>———ChatTool callingUnder evaluation
Try in chat
google/<gemini>google/<gemini>———ChatVisionUnder evaluation
Try in chat
mistral/<large>mistral/<large>———ChatUnder evaluation
Try in chat

Model costs pass through in stablecoins at actual usage. The gateway fee is 5% of model cost (2% with BYOK). Sense Tools are billed separately in DC.

Model and vendor names show compatibility only and imply no partnership, license or endorsement. The gateway's published list is authoritative.

Model access plan

Model familyAccessPlanned
OpenAI GPTOfficial API, within its terms of serviceGateway Beta (Q1 2027)
DeepSeekOfficial API; open weights hosted by inference partnersGateway Beta (Q1 2027)
Meta LlamaHosted open weights under the Llama license, or compliant third-party inferenceGateway Beta (Q1 2027)
QwenOfficial API; hosted open weightsQ2 2027
Claude, Gemini, Mistral and othersOfficial API or BYOK, evaluated case by case under commercial agreementsFrom Q2 2027