Interfaze (YC P26)

@interfaze_ai x @andcolegal Wipe your phone btw it’s a bit dusty

hosting a dinner with @rhobusiness and @interfaze_ai > seed - series b CTOs & HOEs (heads of engineering) > bussin food n drinks 🍷 > july 26, san francisco want a spot? apply below 👇

Inception

Since everyone is talking about GPT 5.6 Sol for object detection while paying $30-$45 for output. we have been doing it 10x better for 10x cheaper at $3.50 for output

Chai night at our office was an amazing night It's basically an excuse for me to shamelessly drink unlimited chai

@yoeven @khurdula @NehallAgrawal That's real chai

We're launching a full public demo right on the landing page Try OCR, Web extraction, Audio transcription and more Make as many requests as you like

To my fellow Chai lovers We're hosting cozy Chai Night for founders and builders Pull up (limited spots)

AIE was crazy this year, 6x bigger than last year but still well curated There were puppies. And a face matching AI that things I look like Ranbir Kapoor

Weights: huggingface.co/interfaze-ai/d…

Try it here: huggingface.co/spaces/interfa…

The First Open Source Diffusion ASR Audio Model 15x faster than whisper We achieved this by fusing DiffusionGemma and Whisper small encoders by denoising an entire transcript in parallel

Full benchmarks: interfaze.ai/leaderboards

What was surprising is that it dropped in GPQA compared to the previous 4.6 model

The latest Claude Sonnet 5 benchmarks are out and here are the results: While it's a huge improvement over 4.6 it still underperforms across all 9 benchmarks for OCR, Object detection, text-to-sql and structured output while maintaining it's high cost

Mistral OCR 4 comes it 4th on olmOCR and Reducto 11th A better perspective where the OCR market is on an open source industry benchmark