Refreshing the AI model selection (04/2025)

Hey RemNote Team,

I wanted to share some thoughts on updating the AI model options in RemNote. As I get into more complex subjects, having access to smarter AI models is incredibly valuable. I’ve compiled a list of newer models that I think would be great additions. To keep the user interface clean, I’d also suggest adding a settings option where users can enable or disable specific AI models in the dropdown menu. This way, we can customize which models we want to use without cluttering the UI.

Here’s my updated list of AI model suggestions:

OpenAI:

- o4 mini (High) - A powerful new model for especially challenging subjects.

- o4 mini - Another solid option for hard subjects.

- GPT 4.1 - Could replace GPT 4o or be added alongside it.

- GPT 4.1 Mini - A potential replacement for GPT 4o Mini with better performance.

- GPT 4.1 Nano - A low-cost option that could use minimal or no credits for lighter tasks.

(Note: I think o3 might be too expensive to include at this time.)

Google:

- Gemini 2.5 Pro - A top-tier model for especially demanding tasks.

- Gemini 2.5 Flash (Thinking) - A strong choice for harder tasks with thoughtful responses.

- Gemini 2.5 Flash - Suggested as a replacement for Gemini 2.0 Flash for better performance.

XAI:

- Grok 3 - A new model for harder tasks.

- Grok 3 Mini (Thinking) - A new model for hard tasks with thoughtful responses.

Meta:

- Llama 4 Maverick - A reliable and affordable vision-capable model for versatile use.

- Llama 4 Scout - A great option for simpler tasks with cost in mind.

Anthropic:

- All models seem up to date. No changes needed here. 👍

Deepseek:

- Deepseek R1 (Thinking) - A very cost-effective, high-performance model (non-vision capable).

- Deepseek V3.1 - Another budget-friendly, high-performing option (also non-vision capable).

Let me know what you think, and thanks for considering this feedback!