Other: Trinity Mini

A 26B sparse mixture-of-experts model from Arcee AI with only 3B active parameters per token, tuned for multi-turn conversations, tool use, and structured outputs in cloud or on-prem deployments (December 2025).

Per Arcee AI:

Trinity Mini is tuned for reasoning, but in testing, it uses a similar total token count to competitive instruction-tuned models. Trinity Mini is trained on 10T tokens gathered and curated through a key partnership with Datology.

Key Features

• Parameters: 26B total, 3B active per token
• Experts: 128 total, 8 active, 1 shared
• Context Window: 128K tokens

Capabilities

• Agent Reliability: Accurate function selection, valid parameters, schema-true JSON, graceful error recovery
• Multi-Turn Conversation: Maintains goals and constraints over long sessions
• Structured Outputs: JSON schema adherence, native function calling and tool orchestration
• Efficient Attention: Lower cost for long-context inference

Use Cases

• Customer-facing applications and agent backends
• High-throughput services in cloud or VPC
• Multi-agent systems and tool orchestration

Links

• HuggingFace
• GGUF Weights
• Playground
• API Documentation
• OpenRouter