What
The Gemini provider in crates/ember-llm/src/gemini.rs may not fully support streaming responses. Add or verify streaming support using Gemini's SSE API.
Why
Streaming makes chat feel responsive — users see tokens as they arrive instead of waiting for the full response.
How
- Check
crates/ember-llm/src/gemini.rs for the current streaming implementation
- Compare with
crates/ember-llm/src/openai.rs which has working streaming
- Gemini uses
streamGenerateContent endpoint with SSE
- Implement or fix the streaming path
- Add a test using
MockProvider if possible
Reference
What
The Gemini provider in
crates/ember-llm/src/gemini.rsmay not fully support streaming responses. Add or verify streaming support using Gemini's SSE API.Why
Streaming makes chat feel responsive — users see tokens as they arrive instead of waiting for the full response.
How
crates/ember-llm/src/gemini.rsfor the current streaming implementationcrates/ember-llm/src/openai.rswhich has working streamingstreamGenerateContentendpoint with SSEMockProviderif possibleReference
crates/ember-llm/src/streaming.rshas the streaming utilities