模型介绍
This model is the first Polish-focused Vision-Language Model (VLM), created by extending the open-source LLaVA architecture with the Bielik language model. Our pipeline integrates high-quality multimodal instruction tuning with Bielik’s strong Polish linguistic abilities, resulting in a VLM that demonstrates significantly improved understanding of Polish language, culture, and context-specific visual reasoning. The model is intended for research and development purposes, specifically focusing on multimodal tasks requiring the Polish language and cultural context. It can be used directly for: This model can be fine-tuned or integrated into larger applications to support specific use cases, such as: * **Accessibility Tools:** Creating applications that describe surroundings or digital content to visually impaired Polish speakers. * **E-commerce:** Generating automated product descriptions based on images for Polish marketplaces. * **Educational Assistants:** Developing tutoring systems that can explain visual content (diagrams, historical photos) to students in Polish. * **Specialized Fine-tuning:** The model can be further fine-tuned on domain-specific datasets (e.g., Polish medical imaging reports or legal document analysis) to improve performance in niche sectors.
主要特点
特点摘要待依据来源补齐，不凭模型名称推断能力。
局限与注意事项
暂无足够来源支持具体缺点结论。请重点验证实际任务表现、时延、数据政策、许可与版本生命周期。
来源报价与费用分析
同一模型在不同渠道、地区和配置下可能采用不同价格。下表独立保留各渠道报价，不将渠道价格视为厂商官方统一价格。
| 来源 / 渠道 | 输入 | 输出 | 计费单位 | 采集时间 |
|---|---|---|---|---|
| 暂未取得可核验报价，请查阅来源渠道。 | ||||
按月估算基础 Token 费用
预估不包含缓存优惠、阶梯定价、多模态附加费用、网络检索、税费及推理额外输出。
公开评测与运行配置
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。
查看完整已收录榜单 →常见问题
开放权重是否代表可以免费商用？
不一定。应核对具体版本许可证、商用门槛和使用政策。自部署还会产生算力和维护成本。
这些价格是实时结算价格吗？
不是实时结算价格，而是带采集时间的来源报价快照。请到对应渠道核对当前版本、区域、折扣和计费方式。
资料来源与可追溯记录
许可证：cc-by-nc-4.0。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。