# d1-2B

> d1-2B is a decision model for text, pictures, and video, built on LFM3-VL-2B. It returns calibrated answers in one forward pass, without generating tokens.

- **Organisation:** LiquidAI
- **Category:** Vision-Language Models
- **Parameters:** 1.9B
- **Context length:** 128,000 tokens
- **Licence:** lfm1.0 (Commercial use with conditions)
- **Released:** 2026-10-24
- **Hugging Face:** 17.1K downloads (30 days), 352 likes
- **Ollama:** `ollama run d1:2b`

## Estimated memory

| Quant | ≈ GB |
|---|---|
| Q2_K | 1.49 |
| Q3_K_M | 1.79 |
| Q4_K_M | 2.01 |
| Q5_K_M | 2.21 |
| Q6_K | 2.42 |
| Q8_0 | 2.87 |
| F16 | 4.63 |

## Runs with

- [llama.cpp](https://selfhost.directory/project/llamacpp)
- [LM Studio](https://lmstudio.ai)
- [MLX](https://github.com/ml-explore/mlx-lm)
- [Ollama](https://selfhost.directory/project/ollama)
- [Transformers](https://github.com/huggingface/transformers)
- [Jan](https://jan.ai) (community)
- [KoboldCpp](https://selfhost.directory/project/koboldcpp) (community)
- [LocalAI](https://selfhost.directory/project/localai) (community)
- [ONNX Runtime](https://onnxruntime.ai) (community)

## About

d1-2B is a decision model for text, pictures, and video, built on LFM3-VL-2B. It returns calibrated answers in one forward pass, without generating tokens.

• Best-in-class: it leads the open d1 models on average and on our board of over a hundred decision benchmarks, ahead of d1-3B at under two thirds of its size.
• Beyond classification: it also extracts text, numbers, and dates with their span, locates objects in pictures, and estimates quantities with an interval.
• Millisecond decisions: 5.7 ms per decision on an NVIDIA H100 and 3.8 ms on an AMD MI325X, with any number of questions about one state in one call.

Find more information about open d1 in our release blog post.

The LFM3 family includes base, text, vision, audio, omni, and decision models. d1-2B decides over text, pictures, and video, and d1-Omni-2B adds audio.

d1-2B is designed for any pipeline step that needs a decision: routing and triage, moderation and guardrails, intent and topic classification, extraction from documents and forms, judging and reranking, and visual inspection. It is not a chat model.

Knowledge-heavy questions favor larger models, and answers get less reliable past about 64K tokens of state. Estimates are reliable on counts and scores, less so on free quantities.

d1-2B runs in Transformers with its own model code. It reads its answers from a single forward pass, so there are no sampling parameters to set.

Transformers quick start:

Values from a document

Pictures and boxes

## Alternatives

- [Qwen/Qwen3.5-2B](https://selfhost.directory/models/qwen/qwen3.5-2b) (2.3B, Apache 2.0, ≈2.22 GB at Q4): Over recent months, we have intensified our focus on developing foundation models that deliver exceptional utility and performance.
- [openbmb/MiniCPM-V-4.7-1B](https://selfhost.directory/models/openbmb/minicpm-v-4.7-1b) (1.6B, Apache 2.0, ≈1.8 GB at Q4): MiniCPM-V 4.7 is the latest model in the MiniCPM-V series, available in a 1B variant. It builds on the family's single-image, multi-image, and video understanding capabilities, combining stronger foundation capabilities with efficient visual encoding and on-device inference.
- [LiquidAI/LFM3-VL-2B](https://selfhost.directory/models/liquidai/lfm3-vl-2b) (1.9B, lfm1.0, ≈2.01 GB at Q4): LFM3-VL-2B is a vision-language model for on-device visual understanding. It takes text, images, and video as input, and combines the LFM3 language model with a new in-house vision encoder.
- [Qwen/Qwen3-VL-2B-Instruct](https://selfhost.directory/models/qwen/qwen3-vl-2b-instruct) (2.1B, Apache 2.0, ≈2.64 GB at Q4): Meet Qwen3-VL — the most powerful vision-language model in the Qwen series to date.
- [Qwen/Qwen3.5-2B-Base](https://selfhost.directory/models/qwen/qwen3.5-2b-base) (2.3B, Apache 2.0, ≈2.22 GB at Q4): Over recent months, we have intensified our focus on developing foundation models that deliver exceptional utility and performance.
- [nvidia/Cosmos-Reason2-2B](https://selfhost.directory/models/nvidia/cosmos-reason2-2b) (2.4B, NVIDIA Open Model License, ≈2.51 GB at Q4): NVIDIA Cosmos Reason 2 is an open, customizable, 2B-parameter reasoning vision language model (VLM) for physical AI and robotics that enables robots and vision AI agents to reason like humans, using prior knowledge, physics understanding and common sense to understand and act in the real world.

More: https://selfhost.directory/models/liquidai/d1-2b/alternatives

## Runs well on

- [GeForce RTX 4060](https://selfhost.directory/models/liquidai/d1-2b/on/rtx-4060): F16, est. 39 tok/s
- [Apple M4 (16 GB)](https://selfhost.directory/models/liquidai/d1-2b/on/apple-m4-16gb): F16, est. 17 tok/s
- [Radeon RX 6700 XT](https://selfhost.directory/models/liquidai/d1-2b/on/rx-6700-xt): F16, est. 55 tok/s
- [GeForce RTX 3060 12GB](https://selfhost.directory/models/liquidai/d1-2b/on/rtx-3060): F16, est. 52 tok/s
- [Arc B580](https://selfhost.directory/models/liquidai/d1-2b/on/arc-b580): F16, est. 65 tok/s
- [Tesla T4](https://selfhost.directory/models/liquidai/d1-2b/on/tesla-t4): F16, est. 46 tok/s

## Similar size

- [Qwen/Qwen3.5-2B](https://selfhost.directory/models/qwen/qwen3.5-2b) (2.3B)
- [Qwen/Qwen3-VL-2B-Instruct](https://selfhost.directory/models/qwen/qwen3-vl-2b-instruct) (2.1B)
- [LiquidAI/LFM3-VL-2B](https://selfhost.directory/models/liquidai/lfm3-vl-2b) (1.9B)
- [openbmb/MiniCPM-V-4.7-1B](https://selfhost.directory/models/openbmb/minicpm-v-4.7-1b) (1.6B)
- [nvidia/Cosmos-Reason2-2B](https://selfhost.directory/models/nvidia/cosmos-reason2-2b) (2.4B)
- [HuggingFaceTB/SmolVLM2-2.2B-Instruct](https://selfhost.directory/models/huggingfacetb/smolvlm2-2.2b-instruct) (2.2B)

## FAQ

**How much VRAM does d1-2B need?**
About 2.0 GB at Q4_K_M with an 8K context. Q8_0 needs about 2.9 GB and full precision about 4.6 GB. Longer contexts need more memory for the KV cache.

**Can d1-2B run on the RTX 4060?**
Yes. It fits at FP16 / BF16 (about 4.3 GB), at an estimated 39 tokens/s. It fits comfortably on the Apple M4 (16 GB), Radeon RX 6700 XT, RTX 3060 12GB.

**Can I use d1-2B commercially?**
Yes, with conditions. It is released under lfm1.0. Custom licence: read the model card before using it commercially.

**What is the context length of d1-2B?**
128,000 tokens (125K). The full window needs far more memory than an 8K context.

**How do I run d1-2B with Ollama or llama.cpp?**
With Ollama: ollama run d1:2b. With llama.cpp: llama-server -hf LiquidAI/d1-2B-GGUF:Q4_K_M, which starts an OpenAI-compatible server.

## Links

- huggingface: https://huggingface.co/LiquidAI/d1-2B
- homepage: https://www.liquid.ai/blog/open-d1
- demo: https://huggingface.co/spaces/multimodalart/jev-decision-index
- ollama: https://ollama.com/library/d1

---
Source: https://selfhost.directory/models/liquidai/d1-2b