Skip to main content

MIT HAN LabImage model

HART-0.7B-1024px

Arena ratingnot rated

n/aUser voteshigher is better

GraduallyBenchnot tested yet

n/aour own rating0 to 100

Priceno price

n/aper imagelower is better

PopularityRank 670 of 875

0out of 100first value

Summary

HART-0.7B-1024px is an image model from MIT HAN Lab. HART is a hybrid autoregressive research model for direct text-to-image generation at 1,024 × 1,024 pixels.

Popularity compared

How often the model is searched for, used and mentioned, compared with every model of its category.

Popularity index

The 100 most popular models of this category, 0 to 100, higher is more popular

20 of 100 models

Stable Diffusion XL 1.072
Anima Base v1.068
SDXL Turbo63
Stable Diffusion Turbo63
Nano Banana 261
FLUX.2 Max55
Krea 2 Raw54
Nano Banana Pro50
FLUX.2 Pro48
MOSS-Image-SFT46
FLUX.2 Flex46
GPT Image 246
LongCat-Image45
MAI-Image-2.645
FLUX.2 Klein 9B45
PixArt-Sigma-XL-2-1024-MS44
Janus-Pro-7B44
HunyuanImage 3.0 Instruct44
MOSS-Image-Base44
HART-0.7B-1024px0

As of November 2026 · Source: gradually.ai popularity index

Over time

First value from November 5, 2026. The timeline appears from the second week.

HART-0.7B-1024px compared

Compare with another model

All specifications

Architecture

Parameters
0.7B
Source: Parameters

Input and output

Maximum output
1,024 × 1,024 pixels
Source: Maximum output

Watermarking and provenance

AI watermarking refers to invisible signals embedded in the output. Visible marks and provenance metadata are listed separately. Unknown means there is insufficient evidence for that particular field.

AI watermarking
Unknown
no source
Watermarking technique
Not known
no source

Licence and access

Weights
Open weights
Source: Weights
Access
Local
Source: Access

Further details

Checkpoint
mit-han-lab/hart-0.7b-1024px
Source: Checkpoint
Local inference
Hybrid autoregressive research model for direct 1,024 × 1,024 text-to-image generation.
Source: Local inference
Additional runtime models
The official example uses Qwen2-VL-1.5B as the text model; the Gradio demo can additionally use ShieldGemma-2B for safety filtering.
Source: Additional runtime models
License
MIT according to the official repository.
Source: License

Sources and data date

TypeEvidence and data date
Additional sourceMIT HAN Lab (retrieved November 1, 2026)
Additional sourceMIT HAN Lab (retrieved November 1, 2026) · MIT HAN Lab · Editorial description reviewed November 1, 2026
Additional sourceOpenRouter: Popularity index (derived), CC BY 4.0
Additional sourceVercel AI Gateway: Popularity index (derived), CC BY 4.0
Additional sourceOllama: cumulative downloads for the popularity index
Additional sourceReplicate: cumulative runs for the popularity index
Additional sourceGoogle Trending Now: limited feed selection for Germany and the US
Additional sourceCreative Commons Attribution 4.0