Not another Jev. TypeLLM adds type-safe generation to LLMs — more types (string, number, boolean, enum), vision, dependent fields and thinking. Playground is open to all with $5 credit. API is rolling out to early-access users in the coming days.

Pricing: • $0.05 / 1M input tokens • $0.50 / 1M thinking tokens (only for fields that think) • Typed outputs are free Request API access via

How is this different from json schema structured output? Faster? Auto retry until fit? Special saucy prompt to help it think hard about types? It's a bolt on for existing models? Too many releases my brain hurts, pls help.

Type safety for LLMs: so now the model can fail to compile instead of confidently making things up. Honestly, progress.

how do dependent fields work? can one answer change the options for the next one?

Currently, the options are fixed; only the next decision depends on the last decision, see for an example. You can write an outer loop to change the options based on the previous decision.

dependent fields are the bit i keep wanting in tool schemas. fewer strings to validate after the call

Time to play with typellm. Do you guys have a discord we can join?

Perfect, that would be a better fit in my case.

Looking forward to your use case! Keen to share it with the community.

This model supports multimodal processing and can handle images, I like it.

I like it!