Eagle
NVIDIAEagle is NVIDIA Research's family of vision-language models, built around data-centric post-training recipes. Eagle 2.5, the current release, pairs a SigLIP2 vision encoder with Qwen2.5-7B-Instruct and is trained for long-context input: up to 512 video frames, high-resolution images and multi-page documents. Eagle models also serve as the vision-language backbone of NVIDIA's GR00T robot models.
The nvidia Hub account also publishes Eagle3 checkpoints; those are speculative-decoding draft models for other LLMs, a different project, and are not part of this entry.
Openness
2 high confidence- weights
- open (Eagle2.5-8B safetensors on the Hub behind an automatically approved access agreement)
- data
- described (open-source video and document sets plus the self-curated Eagle-Video-110K, which is not released)
- code
- partial (stage-2 training scripts and documentation; the data recipe ships only as a debug example)
- license
- NVIDIA-License (Eagle 2.5 weights; non-commercial academic or non-profit research only)
- eagle-2-weights
- CC-BY-NC-4.0 (Eagle2-1B, 2B and 9B, the earlier release)
- repository-code
- Apache-2.0 (repository code)
Eagle 2.5's weights carry an NVIDIA license that allows only non-commercial academic or non-profit research use. The training code is Apache-2.0, and the earlier Eagle 2 models are CC BY-NC 4.0, so no model in the line is cleared for commercial use.
- https://huggingface.co/api/datasets?author=nvidia&search=eagle&limit=50 recorded 2026-11-12
No nvidia dataset on the Hub matches eagle.
- https://huggingface.co/api/models?author=nvidia&search=eagle&limit=100&expand[]=downloads&expand[]=cardData&expand[]=gated&expand[]=createdAt recorded 2026-11-12
Eagle2-1B, 2B and 9B tagged cc-by-nc-4.0 and gated false; Eagle2.5-8B tagged other (nsclv1).
- https://huggingface.co/api/models/nvidia/Eagle2.5-8B recorded 2026-11-12
Hub record for nvidia/Eagle2.5-8B: gated "auto", license other with license_name nsclv1, four safetensors shards.
- https://raw.githubusercontent.com/NVlabs/Eagle/main/Eagle2_5/LICENSE_MODEL recorded 2026-11-12
NVIDIA License, 3.3 Use Limitation: "The Work and any derivative works thereof only may be used or intended for use non-commercially"; "non-commercially" means "for academic or non-profit research purposes only".
- https://raw.githubusercontent.com/NVlabs/Eagle/main/Eagle2_5/README.md recorded 2026-11-12
Eagle 2.5 README: "Combines open-source data (human-annotated and synthetic) with the self-curated Eagle-Video-110K dataset"; training via shell/train_stage2.sh; the data guide uses playground/sft_recipe with a debug recipe.
- https://raw.githubusercontent.com/NVlabs/Eagle/main/README.md recorded 2026-11-12
README: "The code is released under the Apache 2.0 license"; "The pretrained model weights are released under either the CC BY-NC 4.0 License or the NVIDIA License. The models are research preview intended for non-commercial use only."
Adoption
2 high confidenceHugging Face downloads over the trailing 30 days, summed across Eagle2.5-8B and the three Eagle 2 sizes, most of them for Eagle2-2B and Eagle2.5-8B.
- https://huggingface.co/api/models?author=nvidia&search=eagle&limit=100&expand[]=downloads&expand[]=cardData&expand[]=gated&expand[]=createdAt recorded 2026-11-12
Eagle2-2B 13,815, Eagle2.5-8B 10,648, Eagle2-1B 786 and Eagle2-9B 275 downloads in the trailing 30 days (25,524 together).
Capability
3 high confidenceEagle 2.5 reads documents, charts and long videos, level with MiniCPM-V. It does not act in a live interface or take audio input.
- https://raw.githubusercontent.com/NVlabs/Eagle/main/Eagle2_5/README.md recorded 2026-11-12
Eagle 2.5 README: "Eagle 2.5 supports up to 512 video frames and is trained jointly on image + video data"; supported inputs include multi-page documents; Video-MME 72.4 without subtitles, DocVQA 94.1 and ChartQA 87.5 for Eagle2.5-8B.
Verified 2026-11-12