Artificial Analysis ranks HiDream-O1-Image-Dev-2604 as top open weights model

Jun 1, 2026

Independent benchmark firm Artificial Analysis used its blind human preference arena to rank HiDream-O1-Image-Dev-2604 as the top open-weights text-to-image model. It surpassed previous leaders from Alibaba and Black Forest Labs with an ELO score of 1,191. This version builds on the HiDream-O1-Image architecture by adding a prompt-enhancement pipeline previously tested under the pseudonym Peanut.

Arena ELO: 1,191
Open Weights Rank: Number 1
Model Scale: 8B parameters
License: MIT License
Max Resolution: 2,048 x 2,048

The ranking highlights a shift toward unified architectures that eliminate the VAE (a component translating pixels to AI data). By processing pixels and text in one token space, the 8B parameter model matches proprietary quality. It follows a rapid climb for the family, which recently debuted as a top four open source model before this latest fine-tune claimed the lead.

Model weights and the inference pipeline are available on GitHub under the MIT license. While HiDream-O1-Image-Dev-2604 is for local use, the base model and distilled variants are available on the fal API starting at $5 per 1,000 images. These tools support high-resolution 2K generation and instruction-based editing in open-weights workflows without requiring external text encoders.

View the full update on github.com

Artificial Analysis

@ArtificialAnlys3d ago

HiDream-O1-Image-Dev-2604 debuts as the leading open weights Text to Image model in the Artificial Analysis Image Arena, with the base HiDream-O1-Image and HiDream-O1-Image-Dev also available open weights but landing lower on the leaderboard @HiDream_AI's O1-Image family spans three models: the 8B HiDream-O1-Image, its distilled HiDream-O1-Image-Dev, and HiDream-O1-Image-Dev-2604, a fine-tune of Dev with a prompt-enhancement pipeline, previously listed pseudonymously as Peanut. The base and Dev models accept text plus up to 10 image inputs, spanning generation and instruction-based image editing. On the Artificial Analysis Text to Image Arena, HiDream-O1-Image-Dev-2604 leads all open weights models, delivering quality similar to proprietary models like ByteDance's Seedream 4.0 and Black Forest Labs' FLUX.2 [max]. In Image Editing, HiDream-O1-Image is the second-highest open weights model, behind only Tencent's HunyuanImage 3.0 Instruct. Weights and the full inference pipeline (including HiDream's prompt refiner used during evaluation for HiDream-O1-Image-Dev-2604) are open-source on Hugging Face and GitHub under the MIT license. HiDream-O1-Image and HiDream-O1-Image-Dev are also available across third-party API providers including Fal, priced on Fal at $10/1k images and $5/1k images respectively. Congratulations to @HiDream_ai on the releases! See below for comparisons between the HiDream-O1-Image family and other leading models in the Artificial Analysis Image Arena 🧵

115

View on X

Every HeadsUpAI update is written based on its original source and reviewed before it's published. Read our editorial standards →

Keep reading

HiDream-01-Image Ranks as Top Four Open Source Model in Arena

HiDream-01-Image debuted at #27 overall on the Arena.ai Text-to-Image leaderboard, securing the #4 spot among open-source models. The ranking validates the performance of its unified transformer architecture against proprietary systems from OpenAI and Google.

NVIDIA Cosmos 3 takes top open weights rank with agentic reasoning

Artificial Analysis2d ago

NVIDIA Cosmos 3 takes top open weights rank with agentic reasoning

NVIDIA's Cosmos 3 Super models have reached #1 on the Artificial Analysis open-weights leaderboards for both image and video generation. The system uses a reasoning-based architecture to refine prompts before generating high-fidelity visual content.

falMay 10

fal Launches HiDream-O1-Image to Unify 2K Generation and Subject Consistency

fal launched HiDream-O1-Image, a model that handles text-to-image, editing, and subject personalization within a single architecture. By processing raw pixels and text in a unified token space, the model eliminates the need for external components to maintain character consistency or render complex layouts.

Arena.ai Ranks Kimi K2.6 as Top Open Model for Vision and Documents

Arena.aiApr 24

Arena.ai Ranks Kimi K2.6 as Top Open Model for Vision and Documents

Arena.ai confirmed that Moonshot AI's Kimi K2.6 is now the highest-ranked open model in both the Vision and Document Arenas. The model broke into the top 10 overall for document analysis, matching the performance of proprietary frontier models like Gemini 3.1 Pro.