Bilibili Open-Sources Index-Translate, a Qwen3.5-Based Translation Model Family for 150 Languages, Pandaily reports
Reported by Pandaily · not yet confirmed by the company or a second independent outlet. We update this page when it is.
Bilibili's Index LLM team released Index-Translate under Apache-2.0: 2B, 9B and 35B-A3B preview text models for 150 languages, plus dubbing, syllable-control and long-document variants, Pandaily reported.
China context
- Original name
- 哔哩哔哩
- Outside China
- Open weights · pandaily.com
- Claims
- Company-reported; not yet independently evaluated
- For builders
- Developers can integrate a permissively licensed translation model supporting 150 languages into their applications without API costs.
- For investors
- Bilibili's move into open-source AI models may signal a strategic expansion beyond video streaming, potentially creating new monetization or ecosystem plays.
Bilibili's Index LLM team has open-sourced Index-Translate, a family of translation models based on Qwen3.5, under the Apache-2.0 license. The release includes 2B, 9B, and 35B-A3B preview text models supporting 150 languages, along with specialized variants for dubbing, syllable control, and long-document translation.
The model family includes a 35B-A3B mixture-of-experts architecture, suggesting a focus on efficiency for large-scale translation tasks. The inclusion of syllable-control and dubbing variants indicates capabilities beyond text translation, targeting speech-related applications.
Developers outside China gain immediate access to a permissively licensed, 150-language translation model family, reducing the cost and complexity of building multilingual applications without relying on proprietary APIs.
Open-sourcing under Apache-2.0 allows commercial use without licensing fees, potentially lowering barriers for startups and enterprises needing multilingual translation capabilities.
The next verifiable signal is whether Bilibili releases the full (non-preview) versions of the models and whether independent benchmarks confirm the claimed language coverage and translation quality.