Checking DGX service health?
Last checked: waiting for first probe

AI-Powered Mental Health
& Coding Intelligence

Enterprise-grade AI models for mental health support, vision analysis, and code generation. Powered by NVIDIA DGX Spark.

Explore Models
8+
AI Models
128GB
GPU Memory
78.6%
MH Benchmark
24/7
Availability

Available Models

Choose the AI that fits your needs

💙
Checking

Bygheart V5

Mental health support companion with empathetic conversation, crisis detection, and coping strategies.

7B params 8GB Port 8000
Open Chat →
👁️
Checking

Bygheart Vision V2

Visual mental health analysis with facial expression detection, body language interpretation, and real-time video.

Vision 7B 16GB Port 8001
Open Vision →
💻
Checking

Bygheart Coder V3

Current default coder route for chat and agent cutover testing. Benchmarked on DGX with 28.7% HumanEval and 73.2% MBPP.

Coder 7B 15GB Port 8008
Open Coder →
?
Pro Checking

Gemma 4 31B

Premium OpenAI-compatible large model on the DGX for higher-end reasoning, analysis, and long-form prompting.

31B params 19GB /v1 chat
Open Gemma ?
🏛️
Checking

Bygheart COBOL V2

Enterprise COBOL modernization specialist. Translate legacy code to modern languages with semantic preservation.

COBOL 7B 15GB Port 8004
Open COBOL →
Checking

Bygheart Coder V1

Original coder model. Stable and battle-tested for production code generation tasks.

Coder 7B 15GB Port 8005
Open Coder V1 →
🤖
Checking

Coder Agent

ReAct-style autonomous coding agent with tool-use capabilities. Now bound to Coder V3 by default.

Agent Tool-Use Port 8006
Open Agent →

Edge Deployment Ready

GGUF quantized models for local inference

Model F16 Q8_0 Q5_K_M Q4_K_M
Vision V2
Coder V1
Coder V3 (current default) ✓ Default

Q5_K_M (~5GB) recommended for laptops and desktops with 8GB+ RAM