arcee-ai / Trinity-Nano-Base is supported on DeployPad with stability and performance fixes applied. A previously observed output instability issue (nonsensical or “gibberish” responses under certain runtimes) has been resolved. The model can now be run reliably in production configurations.
Feedback from users is requested to validate behavior across workloads and hardware.
Supported Hardware
GPU
NVIDIA H100
NVIDIA H200
NVIDIA L40S
NVIDIA RTX Pro 6000 (Blackwell)
NVIDIA A100
Supported Configurations
Capability
Status
Precision
BF16
Streaming Inference
Supported
OpenAI-compatible API
Supported
Dynamic Batching
Supported
GPU Sharing / Multi-tenant
Supported
Production Deployment
Supported
Stability Notes
Item
Details
Previous Issue
Output instability resulting in nonsensical token generation