TypeSafe AI Releases Jev 1.14 With Images and 262k Context
TypeSafe AI has launched Jev 1.14 under jev-preview, introducing experimental image input, Box bounding-boxes, and a 262k token context at $0.042 per million input tokens.
Stock photo for illustration only, not from the actual event
- TypeSafe AI released Jev 1.14 under the jev-preview alias for System One.
- Added experimental image input and Box bounding-box types for localization.
- Introduced an opt-in 262k context window priced at $0.042 per million tokens.
TypeSafe AI has officially rolled out its latest artificial intelligence update, Jev 1.14, released under the jev-preview alias. This update enhances the System One model by introducing advanced features designed to handle complex multimodal datasets and extended informational inputs for developers and technical users.
A major highlight of this version is the inclusion of experimental image input capabilities, allowing the model to process and interpret visual elements. Alongside this, the update incorporates a Box bounding-box data type, which improves spatial awareness and object localization within images.
Stock photo for illustration only, not from the actual event
In addition to multimodal enhancements, Jev 1.14 introduces an opt-in context window scaling up to 262,000 tokens. This expansion empowers the model to ingest and analyze massive codebases, lengthy research papers, or extensive enterprise documentation in a single session without losing coherence.
The integration of a 262k context window in TypeSafe AI's System One model highlights the industry-wide shift toward long-context processing. This capability is crucial for advanced use cases such as repository-level code analysis and deep document synthesis, allowing AI systems to maintain cross-reference accuracy across enormous blocks of data.
Regarding pricing structure, TypeSafe AI has priced the input tokens for this version at $0.042 per million input tokens, offering a cost-effective solution for developers looking to leverage expanded context and multimodal processing in their applications.
Source: MarkTechPost
Found something wrong in this article? Report an issue with this article
Comments
Leave a Comment