whatmodel.sh / image / Janus-1.3B

Janus-1.3B

+ contribute
Update model informationAdd a place to run itAdd a test resultPropose a benchmarkCorrect a benchmark scoreEdit data on GitHub

Janus-1.3B is an open-weights image model from DeepSeek, released October 2024. Its published size is 1.3B parameters. Licence: deepseek.

Model information sources

Release date ↗ · Weights licence ↗ · Parameter count ↗

1.3B = LLM parameters (DeepSeek-LLM 1.3B base) as the Janus paper and card state them; vision encoder, VQ tokenizer and adaptors are not included. DeepSeek's later Janus-Pro report lists Janus at 1.5B LLM parameters instead.

31.7±3.8#130 of 144 image modelsfrom 3 benchmark results across 2 sources

Score within the LMArena text-to-image comparison set only.

$ how it compares

The image models ranked either side of Janus-1.3B, out of 197 with a score, then more from DeepSeek.

#modelscorewhereRAMreleased
174VILA-U 7B (256px) MIT Han Lab / NVIDIA32±7local~20GB2024-10
175LCM-SSD-1B latent-consistency32±4local~3.6GB2023-11
176Janus-1.3B DeepSeek32±4local~5.5GB2024-10
177IF-M DeepFloyd31±3local~14GB2023-04
178Dreamlike Photoreal 2.0 Dreamlike.art30±3local~2.1GB2023-01
more from DeepSeek
153Janus-Pro-7B DeepSeek38±2api $0.0020/image +local~18GB2025-01
158Janus-Pro-1B DeepSeek36±4api $0.0005/image +local~5.5GB2025-01
162JanusFlow-1.3B DeepSeek36±6comparison only2024-11

see all image models · find the ones that fit your machine

$ what it's good at

text to image▂▅▅▃▄▄█▃▁▂▁▁ #117 of 126 · 94% of its score · 3 tests

Beats VQ-Diffusion (ITHQ) here, though they rank higher overall. Trails CogView2, Versatile Diffusion v1.0, Dreamlike Photoreal 2.0 here, though they rank lower overall.

testresultvs the others in its rundifficultycountswhy
GenEval lab +161.0% ✎▇▃▅▇▁▄▅▃█▃▃▅ beats 13/39easy · avg 83%▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 35%at its level · small run
DPG-Bench lab79.7% ✎▄·▂·▂▆█▄▂▄▂▂ beats 0/17very easy · avg 89%▇▇▇▇▇▇▇▇▇▇▇▇ 31%at its level · small run
WISE (WiScore overall, GPT-4o judge)23.0% ✎▄·▆▂▆█▄▆▂▄·▄ beats 0/22medium · avg 60%▇▇▇▇▇▇▇▇▇▇▇ 28%at its level

$ where to run it

Other ways to run it

  1. Run it as published · needs ~5.5 GB, NVIDIA only · from github.com/deepseek-ai
    git clone https://github.com/deepseek-ai/Janus && cd Janus && pip install -e . && python generation_inference.py

Weights and documentation: huggingface.co/deepseek-ai ↗

$ how the score was worked out · DeepSeek-AI — Janus-Pro techn… 66%, WISE: A World Knowledge-Infor… 28%

Each result implies a score on its own, after adjusting for how hard its test is and who ran it. The score is their weighted average, with a small pull toward the middle for thinly tested models. Each test's weight is its counts column above; here it's added up by who ran the tests. How it works

sourceresultsshare of the scorelicence
DeepSeek-AI — Janus-Pro technical report · lab-reported2▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 66%CC BY 4.0
WISE: A World Knowledge-Informed Semantic Evaluation for Text-to-Image Generation (arXiv v4)1▇▇▇▇▇▇▇▇ 28%CC-BY-SA-4.0
prior · pull toward the middle▇▇ 6%
weighted average= 31.7

not counted (1)

DeepSeek-AI — Janus technical reportGenEval [correct]61.0%the only model in this source's run of the test, so there is nothing to compare it with there; another run of the test counts

$ improve this page

Update model informationAdd a place to run itAdd a test resultPropose a benchmarkCorrect a benchmark scoreEdit data on GitHub

Use a short GitHub form or edit the data through a pull request. Every number needs a source; here's which ones we can use.

data updated · sources and licences · json