whatmodel.sh / image / Janus-Pro-7B

Janus-Pro-7B

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

Janus-Pro-7B is an open-weights image model from DeepSeek, released January 2025. Its published size is 7B parameters. Licence: deepseek.

Model information sources

Release date ↗ · Weights licence ↗ · Parameter count ↗

7B = LLM parameters (DeepSeek-LLM 7B base) as DeepSeek's Janus-Pro report lists them; the SigLIP-Large understanding encoder, the VQ generation tokenizer and the adaptors/heads are not included.

37.6±2.3#113 of 144 image modelsfrom 6 benchmark results across 5 sources

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

$ how it compares

The image models ranked either side of Janus-Pro-7B, out of 197 with a score, then more from DeepSeek.

#modelscorewhereRAMreleased
151BLIP3-o 8B Salesforce Research38±3local~41GB2025-05
152Stable Cascade Stability AI38±4local~13GB2024-02
153Janus-Pro-7B DeepSeek38±2api $0.0020/image +local~18GB2025-01
154NextStep-1-Large-Edit StepFun37±3comparison only2025-08
155Step1X-Edit v1.0 StepFun37±4comparison only2025-04
more from DeepSeek
158Janus-Pro-1B DeepSeek36±4api $0.0005/image +local~5.5GB2025-01
162JanusFlow-1.3B DeepSeek36±6comparison only2024-11
176Janus-1.3B DeepSeek32±4local~5.5GB2024-10

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

$ what it's good at

text to image▂▅▅▃▄▄█▃▁▂▁▁ #98 of 126 · 98% of its score · 6 tests

Beats Latent Diffusion text2img-large (LDM-KL-8-G, LAION-400M), VQ-Diffusion (ITHQ), Stable Diffusion XL 0.9 (base) here, though they rank higher overall. Trails PixArt-α (PixArt-XL-2-1024-MS), SDXL-Lightning (4-step UNet) here, though they rank lower overall.

testresultvs the others in its rundifficultycountswhy
T2I-ReasonBench (overall)42.3% ✎▅▂█▂▂▂····▂▂ beats 1/13easy · avg 69%▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 45%at its level
GenEval lab80.0% ✎▇▃▅▇▁▄▅▃█▃▃▅ beats 27/39easy · avg 83%▇▇▇▇▇ 12%at its level · small run
MMMG knowledge-image generation (MMMG-Score, average over education levels)13.1% ✎▃··▃·█▃▃▅··▃ beats 4/9hard · avg 24%▇▇▇▇▇ 11%typical
WISE (WiScore overall, GPT-4o judge)35.0% ✎▄·▆▃▆█▄▆▂▄·▄ beats 5/22medium · avg 60%▇▇▇▇ 10%at its level
XTC-Bench generation (Overall Gen., scene-graph node agreement)49.9% ✎▄█·▄·······▄ beats 2/4easy · avg 70%▇▇▇▇ 10%at its level
DPG-Bench lab84.2% ✎▄·▂·▃▆█▄▂▄▂▂ beats 3/17very easy · avg 89%▇▇▇▇ 9.4%small run

$ where to run it

Through an API

api · DeepInfra · $0.0020/image · deepinfra.com ↗ · price sourceUSD per generated image on DeepInfra (model page header '$0.002 / image'; catalogue pricing type image_units at 0.2 cents, not deprecated). Product: DeepInfra inference API, text-to-image. API model: deepseek-ai/Janus-Pro-7B. Settings: default output size (Janus-Pro generates at its native resolution), one image. Price basis: published per-image list rate. Currency: USD. Not a DeepSeek direct tariff (DeepSeek's own API serves no image generation). DeepInfra official model page, observed 2026-09-26.

Other ways to run it

  1. Run it as published · needs ~18.3 GB, NVIDIA only · from github.com/deepseek-ai
    git clone https://github.com/deepseek-ai/Janus && cd Janus && pip install -e . && sed -i.bak 's#deepseek-ai/Janus-1.3B#deepseek-ai/Janus-Pro-7B#; s#"role": "User"#"role": "<|User|>"#; s#"role": "Assistant"#"role": "<|Assistant|>"#' generation_inference.py && python generation_inference.py

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

$ how the score was worked out · T2I-ReasonBench: Benchmarking… 45%, DeepSeek-AI — Janus-Pro techn… 21%, MMMG: A Massive, Multidiscipl… 11% and 2 more

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
T2I-ReasonBench: Benchmarking Reasoning-Informed Text-to-Image Generation (Findings of ACL 2026)1▇▇▇▇▇▇▇▇▇▇▇▇▇ 45%CC-BY-4.0
DeepSeek-AI — Janus-Pro technical report · lab-reported2▇▇▇▇▇▇ 21%CC BY 4.0
MMMG: A Massive, Multidisciplinary, Multi-Tier Generation Benchmark for Text-to-Image Reasoning (arXiv:2506.10963v2)1▇▇▇ 11%CC-BY-SA-4.0
WISE: A World Knowledge-Informed Semantic Evaluation for Text-to-Image Generation (arXiv v4)1▇▇▇ 10%CC-BY-SA-4.0
Beyond Accuracy: Benchmarking Cross-Task Consistency in Unified Multimodal Models (XTC-Bench, arXiv:2604.25072v1)1▇▇▇ 10%CC-BY-4.0
prior · pull toward the middle▇ 2%
weighted average= 37.6

$ 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