# d1-3B

Decision models · checked 2026-10-14 · https://dbggr.dev/models/d1-3b/

d1-3B is an open-weight decision model from Liquid AI (LFM Open License v1.0, 3.1B parameters): it answers typed questions, such as yes or no, pick one or pick several, with probabilities instead of writing text. There's no hosted version: you run it on your own hardware. It scores 49.5 (self-hosted; seconds are the board's estimate) on the independent JevBench Capability Score (board v1.7.46), where Jev, the reference, scores 77.1.

- Made by: Liquid AI
- Weights: Open, LFM Open License v1.0
- Size: 3.1B
- Where: Self-host only
- Images: Yes
- Context: 32,768
- Docs: https://huggingface.co/LiquidAI/d1-3B

## d1-3B price

US dollars per million input tokens. Decision models don't write text, so none of the hosted ones charge for output.

| Where | Input | Output | Notes | Checked |
|---|---|---|---|---|
| Self-host only | Self-host | – | – | [2026-10-14](https://huggingface.co/blog/LiquidAI/open-d1) |

## d1-3B benchmarks

| Benchmark | Result | What it measures | Reported by |
|---|---|---|---|
| JevBench v1.7.46 | 49.5 | Capability Score: accuracy and calibration on typed decisions, with Jev as the reference system | [Independent](https://benchmarkheaven.com/jev-models) |
| JevBench time | 0.18 s | Seconds per decision: measured end to end for hosted APIs, the board's estimate for self-hosted models | [Independent](https://benchmarkheaven.com/jev-models) |
| Reported speed | 8 ms a decision on an NVIDIA RTX 4090, 16 ms on a Jetson AGX Thor | Latency on the maker's own hardware | Liquid AI |

Makers' speeds are measured on their own hardware and aren't comparable with each other or with JevBench's timings.

## Decision models like d1-3B

| Model | Input / M | JevBench | Weights |
|---|---|---|---|
| d1-3B | Self-host | 49.5 | Open |
| [Jev 1.13](https://dbggr.dev/models/jev/) | $0.042 | 77.1 | Closed |
| [Clef](https://dbggr.dev/models/clef/) | $0.24 | 75.3 | Open |
| [d1](https://dbggr.dev/models/d1/) | $0.04 | 74.4 | Closed |
| [OpenAI Decisions (GPT-6 Luna)](https://dbggr.dev/models/openai-decisions/) | $0.10 | 73.5 | Closed |
| [Microsoft-Decision-1](https://dbggr.dev/models/microsoft-decision-1/) | $0.042 | 70.8 | Closed |
| [Clef-flash](https://dbggr.dev/models/clef-flash/) | $0.038 | 63.8 | Open |
| [Clef-omni](https://dbggr.dev/models/clef-omni/) | $0.15 | 58.3 | Open |
| [d1-omni-600M (experimental)](https://dbggr.dev/models/d1-omni-600m/) | Self-host | 32.8 | Open |
| [Laya](https://dbggr.dev/models/laya/) | Free | 32.5 | Open |
| [CLM-8B](https://dbggr.dev/models/clm-8b/) | Self-host | 20.1 | Open |
| [Strands Decider 2B](https://dbggr.dev/models/strands-decider-2b/) | Self-host | – | Open |

## FAQ

### How much does d1-3B cost?

d1-3B has no hosted, per-token price: you download it and run it yourself, so the cost is your own compute.

### Is d1-3B open weight?

Yes. Liquid AI publishes the weights under LFM Open License v1.0, so you can run and fine-tune it yourself.

### How fast is d1-3B?

Liquid AI reports 8 ms a decision on an NVIDIA RTX 4090, 16 ms on a Jetson AGX Thor, measured on its own hardware. JevBench gives 0.18 seconds per decision, its estimate for self-hosting.

### How does d1-3B compare with Jev?

On JevBench's Capability Score (board v1.7.46), d1-3B scores 49.5 against Jev's 77.1. Jev costs $0.042 per million input tokens; d1-3B is self-host only. JevBench is a one-person project that uses Jev as its reference, which favours cheap hosted models.