Clarion for startups
A free license to run our Uzbek speech recognition model on your own server: 12 months or until your first paying customer.
Applications open
- applications received
- 1
- accepted into the program
- 1
What you get
The model itself rather than API credits: run it on your own hardware and pay only for your server.
- The uzasr-large model
- The same model as in our cloud: WER 8.1% on real-world speech, punctuation and word-level timestamps.
- Free license
- 12 months with no per-minute charges and no subscription fee.
- Setup help
- We'll deploy the model on your server together with your team.
- Direct line to the team
- Talk to the model's developers directly. Your issues are handled first.
- New versions
- Model updates as they ship.
Who we're looking for
The program is for teams building products around Uzbek speech.
- Early stage
- From idea to your first paying customers.
- A product with Uzbek speech
- Calls, voice bots, media, education and more.
- Ready to share feedback
- Tell us where the model makes mistakes and what's missing. That's how you pay for the license.
- A visible partnership
- If you like, we'll feature you on our site as a Clarion partner.
Terms
The short version: how long it lasts and what happens after the program.
- 12 months free
- Or until your first paying customer or investment, whichever comes first.
- 60 days' notice
- Before the term ends, we'll email you the pricing. No surprise invoices.
- Graduate discount
- 50% off the first year of a paid plan after the program.
- The model stays with you
- You may not share it with others or sell it as an API.
What you need to run it
No GPU of your own? A cloud rental works. We'll help you choose one.
- GPU
- NVIDIA GeForce RTX 2060 or newer (RTX 3060, 4060, 5060) or data center T4, L4
- VRAM
- 4.6 GB for uzasr-large, 2.7 GB for uzasr-base
- CPU and RAM
- 2 cores and 8 GB per worker
- Driver
- NVIDIA with CUDA 12.8+ and NVIDIA Container Toolkit
- Software
- Docker 29+, Docker Compose 2.24+
How to apply
- 1Sign inWith Google or email. It takes a minute.
- 2Tell us about your productWhat you're building and how you'll use speech recognition.
- 3Get a responseThe confirmation and our answer go to the email you provide.
Application
All fields except website and Telegram are required.
