Skip to content
Metatext Decide — typed decisions for your agents, with a probability on every answerGet early access
Metatext
Sign Up

LiquidAI d1-3B

Liquid AIDecision modelOpen weights

LiquidAI d1-3B by Liquid AI is a decision model: it answers typed questions with a probability for each allowed answer instead of generating text. It scores 38.9 on the Jev Decision Index (rank 85).

Pricing

No hosted price is listed. The weights are open: the cost is your own compute.

Specifications

Provider
Liquid AI
Parameters
3B
Released
Oct 2026
Open weights
Yes
Question types
Choice (one of N options), Score (ordered rubric), Yes / no
Mechanism
Decoder LM with a decision readout
Backbone
LiquidAI/LFM2.5-VL-3B
Jev-compatible request
Yes
Runs on CPU
No
License
lfm1.0
Hugging Face
LiquidAI/d1-3B

Benchmarks

Jev Decision IndexECE 0.047
38.9
Benchmark Heaven JevBench (Capability)
49.5

Jev Decision Index: public 49.0, ECE 0.047, Brier 0.390. Lower ECE means its probabilities can be taken at face value.

Self-reported by the publisher

The model's own claims, often on its own sets or a self-run copy of a public suite. Not independently verified and never ranked alongside measured results.

Jev Decision Index public suite (self-run)Self-reportedDecision Index public balanced skill, edition 0.3 kit (release-v3), self-run
49.0

Speed & latency

Median latency per call (Decision Index)
16 ms

Compare LiquidAI d1-3B with…

LiquidAI d1-3B vs amos-1-large—/1M inLiquidAI d1-3B vs amos-1-medium—/1M inLiquidAI d1-3B vs Celeris-1 Decision$0.04/1M inLiquidAI d1-3B vs ScaleDown Classify-1$0.04/1M inLiquidAI d1-3B vs Cloudflare clef—/1M inLiquidAI d1-3B vs Cloudflare clef-flash—/1M in

FAQ

How much does LiquidAI d1-3B cost?

LiquidAI d1-3B has no hosted price listed; its weights are open, so the cost is the compute you run it on.

What questions can LiquidAI d1-3B answer?

LiquidAI d1-3B answers choice (one of n options), score (ordered rubric), yes / no questions, returning a probability for each allowed answer.

Is LiquidAI d1-3B well calibrated?

On the Jev Decision Index its expected calibration error (ECE) is 0.047; below about 0.05, a stated 80% confidence is right roughly 80% of the time.

Is LiquidAI d1-3B open weight, and can it run on a CPU?

Yes, LiquidAI d1-3B has open weights. It needs a GPU to run at useful speed.

Data as of 2026-10-14. Source: Metatext decision-model catalogue, 2026-10-14 (Jev Decision Index, Benchmark Heaven JevBench, Hugging Face). Pricing is per 1M tokens (USD); confirm with the provider before production use.