GLM-5.3 is live. Run Z.ai's latest model on Friendli Model APIs. Try it today ➜
mlfoundations
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Model Details
Model Provider
mlfoundations
Model Tree
OpenGVLab/InternViT-300M-448px
Input Modalities
Output Modalities
Supported Functionality
| Scale | x-large |
| Parameters (N) | 8B |
| Training tokens (D) | 200B |
| Architecture | InternVL-2.5 (InternVLChatModel) |
| Vision encoder | InternViT-300M-448px |
| Language model | Qwen/Qwen2.5-7B |
| Precision | bfloat16 |
| Training compute | ~20,480 H100 hours |
| Training data | DCVLM-Baseline |
| Data mixture | 10% captioning / 70% visual instruction / 15% text-only / 5% multimodal documents |
Core evaluation suite (33 tasks), compared against FineVision, the previous best open pretraining dataset, trained at the same scale with the same architecture and token budget.
| Method | Gen | Know | OCR | Vision | MTL | Text | Core Avg |
|---|---|---|---|---|---|---|---|
| FineVision | 63.5 | 72.8 | 57.5 | 49.6 | 48.4 | 55.7 | 58.2 |
| DCVLM-Baseline (this model) | 73.0 | 73.0 | 53.4 |
Categories: Gen general understanding · Know knowledge · OCR OCR & charts · Vision vision-centric · MTL multilingual · Text text-only.
The model uses custom modeling code, so trust_remote_code=True is required.
python
import torchfrom transformers import AutoModel, AutoTokenizer path = "mlfoundations/dcvlm-8b-model"model = AutoModel.from_pretrained( path, torch_dtype=torch.bfloat16, trust_remote_code=True, low_cpu_mem_usage=True).eval().cuda()tokenizer = AutoTokenizer.from_pretrained(path, trust_remote_code=True, use_fast=False)Image preprocessing follows the standard InternVL-2.5 dynamic-tiling recipe
(448px tiles, max_dynamic_patch=12).
| Code | github.com/mlfoundations/dcvlm |
| Datasets | DCVLM-Baseline 6.25B · DCVLM-Baseline 200B · DCVLM-Balanced 200B |
| Models | 1B · 2B · 4B · |
bibtex
@article{farina2026datacomp, title={DataComp-VLM: Improved Open Datasets for Vision-Language Models}, author={Farina, Matteo and Udandarao, Vishaal and Nguyen, Thao and Kuzucu, Selim and B{\"o}ther, Maximilian and Hochlehnert, Andreas and Ghosh, Adhiraj and Nezhurina, Marianna and Roth, Karsten and Struber, Joschka and others}, journal={arXiv preprint arXiv:2606.28551}, year={2026}}| 63.5 |
| 56.1 |
| 61.1 |
| 63.6 |