MODEL PROFILE

bioclip-2.5-vith14

imageomics · bioclip-2.5-vith14

BioCLIP 2.5 Huge is a foundation model for biology organismal images. It is trained on TreeOfLife-200M on the basis of a CLIP model (ViT-H/14) pre-trained on LAION-2B. BioCLIP 2.5 Huge yields state-of-the-art performance in recognizing various species and further improves the performance on broader biological visual tasks. BioCLIP 2.5 Huge is trained on an updated version of TreeOfLife-200M that adds 19M more images to the original version (dataset to be updated with new metadata and species embeddings). In addition to the slightly larger training dataset, BioCLIP 2.5 Huge was extended to a ViT-H/14 backbone, in comparison with the ViT-L/14 backbone adopted by BioCLIP 2. The training was based on an updated version of the BioCLIP 2 repository (v2.0.0), where `torchcompile` and `pure_bf16` were used for acceleration. The model achieves new state-of-the-art performance on both species classification and broader biological visual tasks, surpassing BioCLIP 2 by 5.7% and 3.5%, respectively. Especially in FishNet, which requires the model to distinguish different habitats, BioCLIP 2.5 Huge demonstrates a 8.7% performance improvement over BioCLIP 2.

multimodallicense:mit
Context windowNot disclosed
Input → Output待核验 → 待核验
Model weightsNot verified
Profile updated2026/11/2 19:15:42
OVERVIEW

Model overview

BioCLIP 2.5 Huge is a foundation model for biology organismal images. It is trained on TreeOfLife-200M on the basis of a CLIP model (ViT-H/14) pre-trained on LAION-2B. BioCLIP 2.5 Huge yields state-of-the-art performance in recognizing various species and further improves the performance on broader biological visual tasks. BioCLIP 2.5 Huge is trained on an updated version of TreeOfLife-200M that adds 19M more images to the original version (dataset to be updated with new metadata and species embeddings). In addition to the slightly larger training dataset, BioCLIP 2.5 Huge was extended to a ViT-H/14 backbone, in comparison with the ViT-L/14 backbone adopted by BioCLIP 2. The training was based on an updated version of the BioCLIP 2 repository (v2.0.0), where `torchcompile` and `pure_bf16` were used for acceleration. The model achieves new state-of-the-art performance on both species classification and broader biological visual tasks, surpassing BioCLIP 2 by 5.7% and 3.5%, respectively. Especially in FishNet, which requires the model to distinguish different habitats, BioCLIP 2.5 Huge demonstrates a 8.7% performance improvement over BioCLIP 2.

Key features

特点摘要待依据来源补齐,不凭模型名称推断能力。

Limitations & considerations

Sources are insufficient to establish specific limitations. Verify task performance, latency, data policies, licensing and version lifecycle.

PRICING & COST

Source prices & cost analysis

Prices vary by channel, region and configuration. Channel quotes are listed independently and are not treated as a single official provider price.

Source / channelInputOutputBilling unitCollected at
No verified prices yet. Check the source channel.

Estimate monthly base token costs

—

Estimates exclude cache discounts, tiered pricing, multimodal surcharges, web search, taxes and additional reasoning output.

INDEPENDENT BENCHMARKS

Public benchmarks & run configuration

暂无已采集的可核验公开评分。模型未上榜不代表能力较低。

View full indexed rankings →

Frequently asked questions

Do open weights permit free commercial use?

Not necessarily. Check the version license, commercial conditions and usage policies. Self-hosting incurs compute and maintenance costs.

Are these real-time billing prices?

These are timestamped source snapshots, not live billing prices. Check the channel for current versions, regions, discounts and billing methods.

Sources & traceability

许可证:mit。模型介绍与第三方评测均以原来源的具体版本和方法为准。