Zense AI gives you access to the latest models from every major AI provider.

You can let Zack decide, or take control and pick exactly the model you want.

Different models suit different tasks — reasoning-heavy homework vs quick factual lookups vs Chinese-language writing.

Switch models

1. In a chat, open the model selector (the model name near the message box).
2. Stay on Zack for auto-routing, or open Other Models to pick a specific model.
3. If the model supports it, set a Reasoning level.
4. Your choice applies to the messages you send next.

Which should I use?

Zack (recommended) — the default auto-router selects the best model for each message. Great for everyday use and agent tools.

Choose a named model when you have a preference. Available families include:

• OpenAI — GPT-5.6, GPT-5, GPT-5 Mini/Nano
• Anthropic — Claude Opus 4.8, Claude Sonnet 4.6 / 4.5 / 5, Claude Haiku 4.5
• Google — Gemini 3.5 Flash, Gemini 2.5 Pro, Gemini 2.5 Flash
• xAI — Grok 4.5, Grok 4.3
• DeepSeek, Qwen, Kimi, Tencent Hy, Mistral, Meta Llama and more

Some older models (e.g. GPT-4.x, o3, o4-mini, Grok 4) have been retired — switch to a current model when prompted.

Reasoning level

Some models support adjustable Reasoning (how much "thinking" the model does before answering). After picking a model:

1. Open the model selector again (or Tutor settings when configuring a Tutor).
2. Choose a reasoning level — options depend on the model (e.g. None, Low, Medium, High, Max).
3. Higher reasoning can improve hard problems but may take longer; the per-message credit stays the flat model rate.

Not every model shows this control — it appears only when the model supports it.

All models work directly in Zense AI with no VPN needed, even where global AI tools are normally blocked.

Note: Some agent tools (like image generation) and longer reasoning may use more credits — see Credits & plans.