Fast response
Prometheus Spark
A compact model for chat, simple actions, classification, and flows where latency matters.
- Embedded assistants
- Routing and short summaries
- Lightweight automation
Prometheus family
AI models lit for real products.
A model line for assistants, agents, transcription, semantic search, and reasoning workflows with stable names, predictable latency, and a consistent developer experience.
5 models Text Voice Embeddings
Model line
Choose the right flame for every task.
Prometheus separates speed, depth, audio, and semantic representation so every product can use the right model without exposing provider names or internal implementation details.
Efficiency
Prometheus Lite
A balance of cost and quality for high-volume product responses and agents that run all day.
- Operational support
- Content generation
- Repeatable workflows
Reasoning
Prometheus Core
The main model for complex tasks, deep analysis, planning, and multi-step agents.
- Extended reasoning
- Code and analysis
- Context-aware decisions
Audio
Prometheus Echo
Speech transcription and understanding for turning meetings, messages, and calls into actionable data.
- Speech to text
- Notes and minutes
- Call analysis
Semantics
Prometheus Atlas
Vector representations for search, recommendations, context retrieval, and agent memory.
- RAG and memory
- Semantic matching
- Intelligent deduplication
Designed for production
A single interface for routing intelligence.
Prometheus lets teams think in capabilities: speed, depth, voice, or embeddings. Behind the interface, the platform can optimize the engine for each task without changing the public product contract.
Routing by task and budget
Stable names for internal teams
Text, audio, and vector layers
assistant.reply Prometheus Lite
agent.plan Prometheus Core
audio.transcribe Prometheus Echo
context.embed Prometheus Atlas
Listen
Turn voice into useful text for support, sales, and operations.
Retrieve
Find knowledge with embeddings and semantic context.
Reason
Plan, compare, and execute tasks with deeper models.
Respond
Deliver fast interactions with models optimized for product use.
Private access