Yachay

Llama

Llama 4 Scout

109B total · 17B active (MoE)

License
Llama Community
Tune type
LORA
Typical cost
~$80
Cold start
5–15 min

Best for

  • general chat
  • code
  • long context
  • multilingual

License terms

Llama 4 Scout is released under the Llama Community license. Yachay has vetted it for commercial use (allowed).

Naming constraint: Derivative must include 'Llama'. The submit form enforces this.

How Yachay tunes it

We default to LORA on Google Cloud Vertex AI Tuning. Yachay auto-selects the right GPU for this model — typically H100 80 GB (Spot) based on the 220 GB FP16 footprint. Hyperparameters are customisable at submit (epochs, batch size, learning rate).

Released April 5, 2025. Weights are pulled from HuggingFace at job start; expect 5–15 minutes of cold start before tuning begins.

Related models

Same family first, then nearest size across families.

Llama
Llama 3.3 70B
70B · ~$80
Llama
Llama 3.1 8B
8B · ~$10
Llama
Llama 3.2 3B
3B · ~$10

Ready to fine-tune Llama 4 Scout?

Sign in with your Condor umbrella account — from ~$0.15 / M training tokens on production-size jobs. No upfront bond; card on file, charged at completion.

Start a tune →

← Back to catalog · See pricing