A sub-400ms latency design target, high-volume throughput, and pricing that works at scale. Mira is the model for the moments where latency matters more than depth.
Capabilities
Mira is designed to a sub-400ms time-to-first-token target so it feels instant. Build chatbots, support agents, and conversational interfaces where any perceptible lag breaks the experience.
Tagging, routing, moderation, sentiment, intent - Mira handles the small decisions that happen millions of times a day. At $0.20 per million input tokens, the economics hold up at high volume.
Fast enough to live inside an editor or search box, Mira powers the inline AI experiences where response time is the product.
Pricing
Input
For prompts at real-time-chat volume.
Output
For generated responses, at any volume.
Per-token pricing that holds up at high volume, metered in USD.
Choose well
Pick Mira when speed and cost are the product - real-time chat, classification, autocomplete, anything that happens thousands of times per minute.
Pick Vaani when you need more reasoning quality, and the workload isn't latency-critical. Most interactive apps fit Vaani, not Mira.
Pick Kavi when the task is rare, hard, and accuracy-critical.