Unlocking Insights from Visual Data: The rise of Multimodal AI
Most modern large language models (LLMs) now handle both text and images. However, businesses often deal with a wealth of details locked within graphical documents - think charts, pdfs, and even handwritten notes. Extracting valuable data from these unstructured sources has historically been a important challenge. Fortunately, advancements in AI are changing that. The demand for models capable of not just reading but also analyzing and even downloading information from unstructured data is rapidly increasing. This shift is driven by the growing need for deep research and data-driven decision-making.The Power of Open Weights
A key advancement in this space is the emergence of open-weight systems. These allow organizations to move away from proprietary AI models and build solutions tailored to their specific needs. This approach fosters greater control, customization, and innovation. Early indications suggest strong interest from developers eager to explore the possibilities offered by these open platforms.Real-World Capabilities: Beyond basic Recognition
The latest AI models are demonstrating notable capabilities in understanding complex visual information. Consider these examples: Handwritten Note Extraction: AI can now accurately decipher handwritten notes from images, a task previously considered difficult. Doodle Comprehension: Models are even capable of interpreting informal sketches and doodles, offering a surprisingly intuitive understanding. These advancements open doors to automating tasks like data entry, document processing, and knowledge finding.Why This Matters to You
If you’re looking to leverage the power of AI within your organization, consider these benefits: improved Efficiency: Automate the extraction of data from visual documents, saving time and resources. Enhanced Decision-Making: Unlock insights hidden within unstructured data to make more informed choices. Greater Versatility: Open-weight systems provide the freedom to customize AI solutions to your unique requirements. Competitive Advantage: Stay ahead of the curve by adopting cutting-edge AI technology. investing in multimodal AI isn’t just about keeping pace with innovation; it’s about unlocking the full potential of your data and driving tangible business value. As these technologies continue to evolve, the ability to seamlessly integrate visual data into your workflows will become increasingly critical for success.The rise in Deep Research features and other AI-powered analysis has given rise to more models and services looking to simplify that process and read more of the documents businesses actually use.
Canadian AI company Cohere is banking on its models, including a newly released visual model, to make the case that Deep Research features should also be optimized for enterprise use cases.
The company has released Command A Vision, a visual model specifically targeting enterprise use cases, built on the back of its Command A model. The 112 billion parameter model can “unlock valuable insights from visual data, and make highly accurate, data-driven decisions through document optical character recognition (OCR) and image analysis,” the company says.
“Whether it’s interpreting product manuals with complex diagrams or analyzing photographs of real-world scenes for risk detection, Command A Vision excels at tackling the most demanding enterprise vision challenges,” the company said in a blog post.
Unlocking Insights from Visual Data: The Rise of multimodal AI
Most modern large language models (LLMs) now handle both text and images. Though, businesses frequently enough deal with a wealth of information locked within graphical documents - think charts, PDFs, and even handwritten notes.Extracting valuable data from these unstructured sources has historically been a significant challenge.Fortunately, advancements in AI are changing that. The demand for models capable of not just reading but also analyzing and even downloading information from unstructured data is rapidly increasing. This shift is driven by the growing need for deep research and data-driven decision-making.The Shift Towards Open Weights
A key development in this space is the move towards open-weight systems. This allows organizations to move away from relying on closed or proprietary AI models. Offering open access fosters innovation and customization, empowering businesses to tailor solutions to their specific needs.Early indications suggest strong interest from developers eager to explore these possibilities.Real-World Capabilities: Beyond Basic Recognition
The latest AI models are demonstrating impressive capabilities. They can accurately extract information from complex visuals, including handwritten notes. Consider these examples: Users are praising the ability to decipher even messy handwriting. Individuals are finding value in AI that can understand and interpret informal sketches and doodles. These examples highlight a significant leap forward in AI’s ability to understand the nuances of visual communication.Why This Matters to you
If you’re looking to unlock the hidden potential within your organization’s visual data, understanding these advancements is crucial. Here’s how this impacts your work: Improved Efficiency: Automate data extraction from documents,saving valuable time and resources. Enhanced Insights: Discover patterns and trends hidden within charts, graphs, and reports. Better Decision-Making: Base your strategies on a more complete and accurate understanding of your data. Competitive Advantage: Leverage AI to gain insights that your competitors may miss. The ability to seamlessly integrate visual data into your AI workflows is no longer a futuristic concept. It’s a present-day reality, poised to transform how you work and make decisions. By embracing these new capabilities, you can unlock a wealth of knowledge and drive significant value for your organization.Unlocking Insights from Visual Data: The Rise of Multimodal AI
Most modern large language models (LLMs) now handle both text and images. Though, businesses frequently enough deal with a wealth of information locked within graphical documents – think charts, PDFs, and even handwritten notes. Extracting valuable data from these unstructured sources has historically been a significant challenge. Fortunately, advancements in AI are changing that. The demand for models capable of not just reading but also analyzing and even downloading information from unstructured data is rapidly increasing. This capability is becoming crucial for deep research and informed decision-making.The Shift Towards Open Weights
A growing number of organizations are seeking alternatives to closed or proprietary AI models. This desire for greater control and customization is driving interest in open-weight systems. Recently, a new option has emerged, offering enterprises a flexible path forward. Early feedback suggests strong developer interest, particularly in the model’s ability to handle complex visual inputs.real-World Capabilities: beyond Basic Recognition
This new technology isn’t just about identifying objects in an image. It demonstrates impressive accuracy in extracting information from challenging sources,like handwritten notes. Users are already highlighting its practical applications: Handwritten Note Extraction: Accurately converts handwritten text into digital, searchable data. Doodle-Friendly AI: Successfully interprets even rough sketches and diagrams. Data Accessibility: Breaks down barriers to information locked within visual formats.These capabilities empower you to unlock insights previously hidden within your visual data. You can streamline workflows, improve analysis, and gain a competitive edge.Why This Matters for Your Business
Consider the implications for your organization. You can now: Automate Data Entry: Reduce manual effort and errors by automatically extracting data from charts and reports. Enhance Research: Quickly analyze large volumes of visual documents to identify trends and patterns. Improve decision-Making: Gain a more complete understanding of your data, leading to better-informed choices.* Unlock Hidden Value: Transform unstructured visual data into actionable intelligence. This evolution in multimodal AI represents a significant step forward in data accessibility and analysis.By embracing these advancements, you can unlock the full potential of your visual data and drive innovation within your organization.Unlocking Insights from Visual Data: The Rise of Multimodal AI
most modern large language models (LLMs) now handle both text and images. Though, businesses frequently enough deal with a wealth of information locked within graphical documents - think charts, PDFs, and even handwritten notes. Extracting valuable data from these unstructured sources has historically been a significant challenge. Fortunately, advancements in artificial intelligence are changing that. Deep research is driving demand for models capable of not just reading visual data, but also analyzing it and even downloading information directly from it. This capability is becoming increasingly crucial for organizations seeking a competitive edge.The Power of Open Weights
One company is responding to this need by offering a new vision model with an open weights system.This approach allows businesses hesitant to rely on closed or proprietary AI solutions to explore and implement the technology more freely. Early indications suggest strong interest from developers eager to leverage this flexibility. The real power lies in the model’s accuracy. Recent demonstrations showcase its ability to accurately interpret even handwritten notes from images. This opens up exciting possibilities. Imagine effortlessly digitizing and analyzing handwritten meeting notes, instantly extracting key data points from complex charts, or streamlining document processing workflows.Beyond Accuracy: A More Human Connection
The advancements aren’t just about technical precision. Users are also appreciating the model’s ability to handle less-than-perfect inputs. Some are even playfully noting its acceptance of “terrible doodles,” highlighting a more forgiving and user-friendly AI experience. This suggests a shift towards AI that understands and adapts to real-world, imperfect data. Here’s how this technology can benefit you: Automated Data Extraction: Quickly pull key information from PDFs, charts, and images. Improved Document Management: Streamline workflows and reduce manual data entry. Enhanced Analysis: Gain deeper insights from visual data sources. Increased Efficiency: Free up your team to focus on higher-value tasks. As multimodal AI continues to evolve, expect even more innovative applications to emerge, empowering you to unlock the full potential of your visual data. This technology isn’t just about processing information; it’s about transforming how you work and make decisions.This means Command A Vision can read and analyze the most common types of images enterprises need: graphs, charts, diagrams, scanned documents and PDFs.
As it’s built on Command A’s architecture, Command A Vision requires two or fewer GPUs, just like the text model. The vision model also retains the text capabilities of Command A to read words on images and understands at least 23 languages. Cohere said that, unlike other models, Command A Vision reduces the total cost of ownership for enterprises and is fully optimized for retrieval use cases for businesses.
how Cohere is architecting Command A
Cohere said it followed a Llava architecture to build its Command A models, including the visual model.This architecture turns visual features into soft vision tokens,which can be divided into different tiles.
These tiles are passed into the Command A text tower, “a dense, 111B parameters textual LLM,” the company said. “In this manner,a single image consumes up to 3,328 tokens.”
Cohere said it trained the visual model in three stages: vision-language alignment, supervised fine-tuning (SFT) and post-training reinforcement learning with human feedback (RLHF).
“This approach enables the mapping of image encoder features to the language model embedding space,” the company said.“In contrast, during the SFT stage, we simultaneously trained the vision encoder, the vision adapter and the language model on a diverse set of instruction-following multimodal tasks.”
Visualizing enterprise AI
Benchmark tests showed Command A Vision outperforming other models with similar visual capabilities.
Cohere pitted Command A Vision against OpenAI’s GPT 4.1, Meta’s Llama 4 Maverick, Mistral’s Pixtral Large and Mistral Medium 3 in nine benchmark tests. The company did not mention if it tested the model against Mistral’s OCR-focused API, Mistral OCR.
Command A Vision outscored the other models in tests such as ChartQA, OCRBench, AI2D and TextVQA. Command A Vision had an average score of 83.1% compared to GPT 4.1’s 78.6%, Llama 4 Maverick’s 80.5% and the 78.3% from Mistral Medium 3.
Most large language models (LLMs) these days are multimodal, meaning they can generate or understand visual media like photos or videos. though, enterprises generally use more graphical documents such as charts and PDFs, so extracting information from these unstructured data sources frequently enough proves difficult.
With Deep Research on the rise,the importance of bringing in models capable of reading,analyzing and even downloading unstructured data has grown.
Cohere also said it’s offering Command A Vision in an open weights system, in hopes that enterprises looking to move away from closed or proprietary models will start using its products. So far, there is some interest from developers.
