Models overview
Mira is a family of language models optimized for different tasks. This guide helps you compare available models and choose the right one.
Current models
ModelDescriptionContextMax output
miraGeneral-purpose model for most tasks32K4K
mira-proProfessional grade with thinking mode64K8K
mira-maxMaximum capabilities with thinking mode128K16K
The model list is available programmatically via the GET /api/v1/models endpoint. The response includes context_window and max_output_tokens for each model.
Choosing a model
If you're unsure which model to use, start with mira for most tasks. For tasks requiring deep reasoning, use mira-pro or mira-max with thinking mode.
- mira — Best balance of speed and quality. Ideal for chat, content generation, simple coding tasks.
- mira-pro — Professional grade with thinking mode. For enterprise tasks, long documents, complex coding, and deep reasoning.
- mira-max — Maximum capabilities with thinking mode. For the most complex tasks, large codebases, research.
Pricing
ModelInput (1M tokens)Output (1M tokens)
mira50 ₽150 ₽
mira-pro100 ₽300 ₽
mira-max250 ₽750 ₽