Real-time chat
Designed to a 400ms latency target so replies feel immediate: chat interfaces, support agents and anything where a visible pause breaks the experience.
ModelsFast · Mira
Fast. Built to a 400ms latency target.
Mira is the Fast size: designed to a 400ms latency target and priced for volume, for the moments where a quick answer beats a long one. Auto sends short prompts here.
64K
tokens of context, the same on every plan.
2K
tokens of output in one reply, at most.
$0.20
per 1M input tokens through the API.
$0.40
per 1M output tokens through the API.
Text in, text out. Rates are USD per 1M tokens, metered per request. API prices exclude GST. 18% GST is added for India-billed accounts (input credit available). On the web, every plan includes Mira.
Built for
What Mira is for.
High-volume classification
Tagging, routing, moderation, sentiment, intent: the small decisions that happen millions of times a day. At $0.20 per 1M input tokens, the economics hold at volume.
Suggestions and short answers
Quick enough to live inside a search box or an editor through the API, for the inline experiences where response time is the product.
Pick by task
Three sizes. Auto picks one for each message; you can always choose.
PICK IT WHEN
Speed and cost are the product: real-time chat, classification, autocomplete, anything that happens thousands of times a minute.PICK BALANCED WHEN
You need more reasoning and the workload is not latency-critical. Most interactive apps fit Vaani.About VaaniDeep· KaviPICK DEEP WHEN
The task is rare, hard and accuracy-critical, or the document is long.About KaviON EVERY PLAN
Speed is the product.
Free on the web, no card required. Per token through the API, at the rates above.