Cerebras Setup

The fastest AI processing measured - $5 of credits, and the card can go straight back out.

Cerebras runs open-weight models on its own wafer-scale hardware, and it is the fastest AI processing VivaDicta can reach — qwen-3.8-27b cleans up a dictation in about 0.3 seconds, quicker than anything else in the model ranking. Cerebras is an AI processing provider only; transcription stays with whichever model you have selected.

Free Tier

• $5 of credits on signup, which lasts a very long time at dictation-sized requests — each enhancement is roughly a thousand tokens.
• A card is required to issue a key, even though the credits are free. You can delete the card again straight after signing up, so nothing can ever be charged.

Get Your API Key

1. Go to cloud.cerebras.ai and sign up.
2. Add a payment method — the key is not issued without one.
3. Create an API key and copy it.
4. Remove the card from billing settings if you want to be certain it is never charged. The $5 of credits stays.

Set Up in VivaDicta

1. Open Settings → AI Providers in the app.
2. Select Cerebras and paste your API key.
3. Pick the model in the Viva Mode you want it for — qwen-3.8-27b is the default and the one measured above.

Which Model

• qwen-3.8-27b — 9.2 quality at 0.3s. The fastest option measured, and what VivaDicta selects by default.
• gpt-oss-120b — 8.7 at 0.32s. A larger model at essentially the same speed, slightly weaker on dictation cleanup.

Cerebras retired gemma-4-31b from its public endpoints in September 2026. If you had it selected, switch to qwen-3.8-27b.