Project AURIONAI Use-Cases Research & Innovation
ByteDance
multimodal LLM

BAGEL (BAGEL-7B-MoT)

ByteDance Seed unified multimodal model for image understanding, text-to-image generation and image editing, released with open weights.

ByteDanceByteDanceChina
Teaser from the BAGEL repository: example images generated and edited by BAGEL, with the model's understanding and generation tasks.
Levelmodel
Classmultimodal LLM
Parameters14B
Contextnot in the notes
Released20 May 2025
Opennessopen weights permissive
Runs onnot in the notes
Statusactive

Benchmark Results

vendor-reportedindependent
Artificial Analysis Text-to-Image Arena AA-Image-T2I v2.0generative media
Rank range 161 to 163independent
706 95% CI 697 to 715 (+/-9)
5009001,300

Score in Elo; the axis starts at 500, as Elo has no zero. Higher is better.

rank 162 of 168, AA Text-to-Image Arena, 168 models11 Oct 2026 artificialanalysis.ai
MMMUmultimodal
vendor-reported
55.3
050100

Score in %. Higher is better.

BAGEL model cardMay 2025 huggingface.co
Where this comes from
Every score names who ran it: vendor-reported, maintainer-run or independent. A rank is shown with the population it was ranked in (gap rule 3).

Lineage

finetune
Qwen2.5-7B-Instruct to BAGEL (BAGEL-7B-MoT): finetune, high confidenceQwen2.5-7B-InstructQwen2.5-7B-InstructAlibaba (Qwen), Sep 2024BAGEL (BAGEL-7B-MoT)BAGEL (BAGEL-7B-MoT)ByteDance, May 2025
Alibaba (Qwen)Qwen2.5-7B-Instruct is the parent (finetune)high confidencemodel card
Where this comes from
Parent and child relations from model cards, Hugging Face base_model fields and the genealogy tables of the research notes. The tree shows two generations each way, 2 models. Confidence and evidence are the notes' own.

R&D Topics

Omni-modal and unified multimodal modelsexemplar

Identity

Developer
ByteDance
Family
not in the notes
Released
20 May 2025 sourcesinferred from page
Weights published
–
Record checked
11 Oct 2026
Where this comes from
Developer and dates from the research notes. Release date is the vendor's public announcement; weights date is when open weights were published.

Architecture and Size

Architecture
MoE Mixture-of-Transformer-Experts (MoT) with pixel-level and semantic-level image encoders; next group of token prediction
Total parameters
14B sources
Active parameters
7B
Checkpoint parameters
–
Note
Card: 7B active, 14B total; safetensors metadata lists 15B
Context length
not in the notes
Input
text, image, video
Output
text, image
Where this comes from
Parameter counts as the developer states them. Where the checkpoint holds a different total, both are shown (register A16).

Openness and Access

Openness
open weights permissive
Licence
Apache License 2.0 sources
Access routes
open weights download
Where this comes from
The licence file governs over registry tags and aggregator labels.

Deployment

Runs on
not in the notes
Footprint
not in the notes
Runtimes
not in the notes
Where this comes from
Deployment tiers and runtimes as named in the notes. Footprint is the notes' figure, not a measurement by AURION.

Indonesia

Indonesian support
not stated
Regional languages
not in the notes
Jakarta hosting
unknown
Languages
not in the notes
Where this comes from
Indonesia fields are carried on every model; unknown is an allowed value.

Industry Use Cases

No industry use case links this model.

Sources

  1. huggingface.co/ByteDance-Seed/BAGEL-7B-MoT https://huggingface.co/ByteDance-Seed/BAGEL-7B-MoT
  2. artificialanalysis.ai/image/leaderboard/text-to-image https://artificialanalysis.ai/image/leaderboard/text-to-image

To cite this page: Project AURION, AI Use-Cases Research & Innovation, "BAGEL (BAGEL-7B-MoT)", https://aurion.semai.id/models/bytedance-bagel/, records of October 2026.