October 8, 2025
A lightweight multimodal model with advanced reasoning capabilities, 128K context window, and multilingual support across 140+ languages. Now available for early preview for pro users.
Spec-4-mini is the latest lightweight multimodal model from SVECTOR with advanced reasoning and multimodal capabilities, enabling it to achieve strong performance across a variety of text generation and image understanding tasks, including question answering, summarization, and reasoning.
Spec-4-mini models are multimodal, handling text and image input and generating text output, with open weights for both pre-trained variants and instruction-tuned variants. Spec-4-mini has a large, 128K context window and multilingual support in over 140 languages.
Their relatively small size makes it possible to deploy them in environments with limited resources such as laptops, desktops or your own cloud infrastructure, democratizing access to state-of-the-art AI models and helping foster innovation for everyone.
Spec-4-mini strikes a balance between capability and resource efficiency, making it suitable for researchers, developers, and enterprise adoption.
Text + Image reasoning capabilities enable comprehensive understanding across multiple modalities.
Handle extensive documents and maintain coherence across large contexts.
Advanced logical reasoning and programming capabilities for complex problem-solving.
Deploy anywhere with quantization support and SVECTOR's Enterprise Deployment Service (EDS).
Spec-4-mini is first pre-trained with a data recipe that includes publicly available Internet data, data produced by third-parties for SVECTOR, data from users or contractors, and internally generated data.
We perform data filtering procedures on the training data, such as de-duplication and classification, to ensure data quality and safety prior to training.
In addition to pre-training, our recipe uses a variety of reinforcement learning techniques—human feedback, verifiable rewards, and model grading—along with supervised finetuning of specific capabilities.
Large-scale corpora including text and image-text pairs with rigorous filtering and de-duplication.
Supervised learning combined with reinforcement learning from human feedback (RLHF).
Model grading and verifiable rewards to ensure safety, accuracy, and helpfulness.
Because our models push the frontier of AI capabilities, we are committed to mitigating their risks through both evaluating model behaviors and implementing safeguards.
Risk Assessment: With our mitigations, we believe that Spec-4-mini overall presents a low risk for malicious use and loss of control.
Enterprise Deployment as Service (EDS) provides managed infrastructure, security, and support for Spec-4-mini deployment at any scale.
