Complete Compatibility Guide

AI Model Guide

Which model to use for each button/feature — and which to avoid

Quick Tips

•Safest for all features: Gemini 2.5 Flash, Groq GPT-OSS 120B, GitHub GPT-4o/GPT-5
•Market Research only works with Gemini — no other provider
•Do not use 8B models (Llama 3.1 8B, Phi-4 Mini) for Special Modes
•Seasonal + Marketplace combo has the largest system prompt — use large models only
•Collection mode needs cohesion — use 70B+ models
•HALAL compliance: small models (8B/3B) may include human figures — use large models

HALAL Compliance Risk by Model

How reliably each model tier follows the HALAL rule (no human figures/faces/hands)

Flagship (Gemini Pro, GPT-5, Qwen 235B)
99%+
Recommended (120B, GPT-4o, Flash)
98%+
Good (70B, 49B, 32B, 31B)
90%+
Limited (11B-20B)
80-85%
Caution (3B-8B)
50-70%
✓ Gemini 2.5 Pro, Gemini 2.5 Flash, Kimi K2-0905 +12✗ Llama 3.1 8B, Qwen 2.5 VL 7B
✓ Gemini 2.5 Pro, Gemini 2.5 Flash, Kimi K2-0905 +12✗ Llama 3.1 8B, Mistral Nemo 12B, Pixtral 12B +3
✓ Gemini 2.5 Pro, Gemini 2.5 Flash, Kimi K2-0905 +12✗ Llama 3.1 8B, Qwen 2.5 VL 7B
✓ Gemini 2.5 Pro, Gemini 2.5 Flash✗ Kimi K2-0905, GPT-OSS 120B, Llama 4 Maverick +36
✓ Gemini 2.5 Pro, Gemini 2.5 Flash, Kimi K2-0905 +12✗ Llama 3.1 8B, Qwen 2.5 VL 7B
✓ Gemini 2.5 Pro, Gemini 2.5 Flash, Kimi K2-0905 +12✗ Llama 3.1 8B, Qwen 2.5 VL 7B

Best Models by Use Case

Prompt Generation (General)
Gemini 2.5 Flash, GPT-OSS 120B (Groq)
Fast, balanced, free
High Quality Content
Gemini 2.5 Pro, GPT-5 (GitHub)
Flagship models, highest quality
Bulk / Fast Generation
Groq Llama 3.1 8B, Gemini Flash-Lite
Maximum speed, lowest latency
Image Metadata (Vision)
Gemini Flash, Pixtral 12B, GPT-4o
Best at image analysis
Reasoning / Coding
o4-mini (GitHub), DeepSeek R1 (OR), Kimi K2 (Groq)
Thinking models, complex tasks
Multilingual Content
Mistral Nemo, Qwen 3 32B (Groq)
Broad language support