Pinned
Real-time AI (audio/video in, voice out) on an M3 Pro with Gemma E2B
- People on Reddit said that Qwen models are better than Gemma for coding, but Gemma is better for everything else. I ran EQ-Bench 3 for Gemma 4 E4B and Qwen 3.5 4B, and Gemma is better than Qwen: ELO 1275.9 vs 1124.1
- I ran English audio benchmark for Gemma 4 E4B using open_asr_leaderboard setup, and the average WER is 9.67. The result is good but not great. It's still a very interesting model since if you're limited on memory, this is still probably the best model that can do everything. If
- Ok I just tried @Cloudflare Email Sending service since I want to migrate from Resend. Seems like you can only send email via the worker or REST API for now? Can I send via SMTP? I'm using @supabase and it needs the SMTP credentials. I can't find anything on the dashboard or
- Claude code is my favorite UI for FFmpeg
