Prompt exceeds max context length error (LeapGenerationPromptExceedContextLengthException)

[Asim-App]

When I try to send a long prompt, I get the following error:

LeapGenerationPromptExceedContextLengthException
Prompt too long: The prompt exceeds the max context length of 4096 tokens.

Steps to Reproduce:

Load model using Leap SDK.
Pass a prompt longer than 4096 tokens.
Observe the error.

Expected Behavior:
The SDK should:

Either automatically truncate the prompt or provide a clear way to handle longer inputs.
Expose the max context length in the API so developers can validate before sending.

Actual Behavior:
Error is thrown directly without any way to manage/handle the prompt size.

Additional Context:

There’s also no option in SDK to clear prompt history or reset the conversation without unloading the model.
It would be helpful if there’s a method like resetHistory() or truncatePrompt().
Also, please confirm if token size limit is fixed per model or configurable.

Environment:

Leap SDK version: v0.5.0
Platform: Android (Kotlin)
Device: Pixel 9 pro
Model: LFM2-700M

[linmx0130]

Hi @Asim-App, thanks for raising this issue.

1. If you want to reset the history, you can create a new conversation from the model runner. Each conversation has a clean history when it is created.
2. Currently, max content length is fixed in the model bundle configuration. Most bundles on Leap model library have a maximum content length of 4096.

We will work on this feedback and provide more options in configuring the max content length to support more tokens.