Good but…
Response from developer
Thank you for your feedback! If you’re using an On‑Device model, try the smaller Spike 2B model for faster responses. Also enable only the tools you need in the Quick Menu—larger models and more active tools require more processing and can make answers take longer. We’re continuing to improve performance and make these choices clearer in the app.
No Response
Response from developer
Sorry—Spike should not have just sat there. In version 1.6.5, messages could be sent without a configured model source, and the connection error was sometimes hidden instead of explaining what was missing. Spike requires either a configured server or a downloaded On-Device model. We appreciate you bringing this to our attention.
Nothing Loads; Slow AI replies
Response from developer
You’re right—waiting seven minutes for a reply is far too long. This version can run AI entirely on the iPhone, and larger On-Device models may be very slow depending on the device. Selecting a smaller compatible model should improve performance, but the app should have explained that tradeoff much more clearly. Thank you for the feedback.
Tools, MLX, and Voice Chat!
Not Bad
You checked every other box except for that one.. And it would be nice if you also added MCP Tools. Can we make this app more like a mobile agent?
