Why custom models?
Instead of general purpose models which try to do everything, custom models are LLMs tailored to perform a specific task.
Tailored to your use case
Custom models are designed to be great at just one use case, and thus outperform general purpose models.
Cheaper to run
Custom models are smaller, and hence cheaper to run than general purpose models. Get more tokens per dollar without losing quality.
Faster inference times
Custom models can be run at a much higher throughput than massive general purpose models. This enables much better experiences built on top of these models.
Stay in control
Own the weights, integrate with your tech stack, and keep your data private. Our users own the models trained with Glaive, and are free to host them anywhere.
Consistent quality responses
General purpose models can be unpredictable. Custom models are designed to be consistent and reliable.
Schemas instead of prompt engineering
Instead of relying on prompt engineering, Glaive uses schemas to define the structure of your inputs and outputs.
Dataset generation
No need to bring your own data, Glaive generates high quality synthetic data for you.
Rapid iteration
Need to add new information, or change the way your model works? Glaive makes it easy to iterate and improve on your model.
Glaive exclusive
Creating a model with Glaive
1 | Describe task
![Describe task](/_next/image?url=%2F_next%2Fstatic%2Fmedia%2Fdescribe.a8254358.png&w=3840&q=75)
2 | Edit knowledge graph
![Edit knowledge graph](/_next/image?url=%2F_next%2Fstatic%2Fmedia%2Fedit.b0ea57fb.png&w=3840&q=75)
3 | Design schema
![Design schema](/_next/image?url=%2F_next%2Fstatic%2Fmedia%2Fschema.5da170a4.png&w=3840&q=75)
4 | Create dataset
![Create dataset](/_next/image?url=%2F_next%2Fstatic%2Fmedia%2Fcreate.1c228824.png&w=1920&q=75)
5 | Train model
![Train model](/_next/image?url=%2F_next%2Fstatic%2Fmedia%2Ftrain.cbca33d9.png&w=3840&q=75)
6 | Repeat to improve
![Repeat to improve](/_next/image?url=%2F_next%2Fstatic%2Fmedia%2Frepeat.a835d650.png&w=3840&q=75)