TypeLLM @TypeLLM

LLMs with type-safe generation typellm.ai London Joined December 2025

Tweets

103
Followers

2K
Following

57
Likes

29

TypeLLM @TypeLLM

a hour ago

@abhi43210 Hi, we are open source: github.com/TypeLLM/TypeLLM

0 0 0 12 0

View Details

Jev answers every question instantly. That's great, until a question is hard. TypeLLM is System 1 when it can be, System 2 when it has to be. "thinking": "auto" picks per question, per call. Read more: typellm.ai/blog/thinking-…

2 2 24 1K 17

View Details

TypeLLM @TypeLLM

11 hours ago

Try it now in our playground: typellm.ai/examples/lette…

0 0 0 92 0

View Details

TypeLLM @TypeLLM

20 hours ago

@neural_avb This approach to choice-order invariance makes the logit prediction for each choice independent of the other choices.

0 0 0 83 0

View Details

TypeLLM @TypeLLM

23 hours ago

Thinking in TypeLLM is now 3x faster, with no loss in accuracy. Get better typed decisions with faster thinking Try it now: typellm.ai/dashboard/play…

0 1 26 2K 15

View Details

TypeLLM @TypeLLM

a day ago

Hi Solomon. If a ticket is classified as "feature-request", both "severity" and "page_on_call" will be skipped. It is the right behaviour, since "severity" only executes when it is in ["bug", "incident"] and "page_on_call" only executes when "severity" executes. We will add more explanation to this case to make it clearer.

0 0 0 8 0

View Details

TypeLLM @TypeLLM

2 days ago

In Jev, every field is generated independently. In TypeLLM, earlier answers decide which fields come next: classify an email, then extract the amount if it is an invoice, or the date if it is a meeting request. All in one API call, with `when`.

8 4 86 7K 72

View Details

TypeLLM @TypeLLM

2 days ago

@typerca Thanks! RCA is a great use case!

0 0 0 86 0

View Details

TypeRCA @typerca

2 days ago

@TypeLLM this is especially interesting for RCA. investigations are naturally conditional — what you check next depends on what the previous check found. “when” looks like a natural way to express that branching. definitely something we want to test in TypeRCA.

1 1 2 169 0

View Details

TypeLLM @TypeLLM

2 days ago

@ares_mheinke So if it is skipped, which means its condition in the “when” field is not satisfied.

0 0 0 86 0

View Details

TypeLLM @TypeLLM

2 days ago

@Mashimaro_AI Yeah sir!

0 0 0 57 0

View Details

TypeLLM @TypeLLM

2 days ago

Learn more in our blog post typellm.ai/blog/condition…

0 0 0 206 0

View Details

TypeLLM @TypeLLM

2 days ago

Try it now in our playground: typellm.ai/examples/email…

1 0 0 344 1

View Details

TypeLLM @TypeLLM

3 days ago

@rae1101x Hi, here is one example that supports this function: typellm.ai/examples/email…

0 0 1 15 0

View Details

TypeLLM @TypeLLM

6 days ago

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. typellm.ai/dashboard/play…

13 65 528 163K 543

View Details

TypeLLM @TypeLLM

3 days ago

@delliott Does the output of the first layer become the new tokens?

0 0 0 196 0

View Details

TypeLLM @TypeLLM

3 days ago

We need a new kind of foundation model that can serve as a universal probabilistic graphical model—supporting uncertainty quantification and propagation before making a decision. TypeLLM is taking a first step in this direction. In v0.5.0, we already support user-defined conditional graphical models. Uncertainty propagation is coming next. More to come.

Thore Graepel @ThoreG

3 days ago

Ten years ago, AlphaGo’s Move 37 shocked the world. It wasn’t intuition alone that produced it. AlphaGo could search possible futures, test its instincts and reason about what would happen next. In a new piece for @techreview, I argue that today’s most advanced AI systems are

48 113 673 91K 453

0 0 4 569 3

View Details

TypeLLM @TypeLLM

3 days ago

Just set "thinking": "auto": "hello" → no thinking → 0 "strawberry" → thinks → 3 Try it now in the TypeLLM playground typellm.ai/examples/lette…

0 0 1 149 0

View Details

TypeLLM @TypeLLM

3 days ago

System 1 models like Jev answer at once: fast, cheap, right when the answer is obvious. System 2 models reason step by step first: slower, but right when the steps matter. Most models are one or the other. TypeLLM decides per field, on every call.

1 0 3 254 1

View Details

TypeLLM @TypeLLM

3 days ago

How many r's are in "strawberry"? System 1 models like Jev answer in one shot and get it wrong. TypeLLM thinks first and says 3, and still answers "hello" (0) instantly. Two systems, one model. 🧵