Skip to main content
Mantle gives you models from several providers in one place. The list changes as providers release new models, so the model picker in Mantle is the best place to see what is available on your plan. The current hosted catalog spans eight providers: OpenAI, Anthropic, Google, xAI, DeepSeek, Moonshot AI, Z.AI, and Mistral AI. Paid plans can use all 25 currently supported hosted models. Free includes GPT-5 Nano, GPT-5.6 Luna, Gemini 2.5 Flash-Lite, DeepSeek V4 Flash, GLM-5.3-Flash, and Mistral Small 4. Because provider availability changes, the model picker remains the source of truth for exact models and capabilities on your plan.

​
Pick by task, not by brand

New chats start with DeepSeek V4 Flash, selected for its strong quality, speed, and credit efficiency. If that choice cannot be used because it is no longer recognized or allowed on the current plan, Mantle replaces it with GPT-5.6 Luna. Provider failures during a started response remain visible instead of silently switching models, so the model identity, tool behavior, and credit record stay accurate. You can switch models later without losing the conversation.

​
Switch models in the same chat

Select the current model beside the message box, choose another one, and send your next message. The new model receives the available conversation context. This is useful when you want to:
  • get a second opinion;
  • use a fast model for a first draft and a stronger model for review;
  • compare two ways of solving the same problem;
  • move from text work to an image-capable model.
Ask the second model to review the earlier answer against clear criteria. This produces a more useful comparison than asking the same vague question twice.

​
Understand credits

Different models and features use different amounts of credits. In general, larger models, more reasoning, longer answers, tools, and web search can use more. Mantle shows relative credit-use levels in the model picker and on the pricing page. Use an efficient model for routine work and save higher-credit models for jobs where the difference matters. After a run, you can review the credits that were actually used. Conversation history also matters. A long chat may require the model to read more earlier messages each time you ask a new question. Start a fresh chat when you move to a different tender, contract, client, or topic, even if some repeated input can be cached.

​
Models can have different abilities

Not every model supports every feature. The model picker and message box show what is available, including image input, web search, and reasoning controls. If an option is unavailable, choose another model or simplify the request. See Plans and credits for plan access and AI chats for the full chat workflow.