Cohere Inc

Cohere Reaches $7B Valuation with $100M Extension Round, Announces Strategic Partnership with AMD

September 24, 2025 - Enterprise-focused AI model developer Cohere has raised an additional $100 million, boosting its valuation to $7 billion. The funding is an extension of its oversubscribed $500 million round in August, which had previously set the company’s valuation at $6.8 billion, according to a company statement.

Alongside the funding news, Cohere revealed a notable partnership with chipmaker AMD, which is also among its investors. This move comes as rival OpenAI recently secured a commitment of up to $100 billion from Nvidia, the dominant force in the GPU market.

Cohere's full Command-family AI model suite—including vision, translation, and reasoning models—can now operate on AMD’s Instinct GPUs, a direct competitor to Nvidia's hardware. AMD will also implement Cohere’s AI technology internally, although Cohere clarified to TechCrunch that it remains committed to supporting Nvidia GPUs and is not shifting exclusively to AMD.

Founded in 2019 by Aidan Gomez, a co-author of the seminal “transformer” paper that ignited the current generative AI surge, Cohere was initially a standout in the AI landscape. However, its rise has been somewhat eclipsed by OpenAI and Anthropic, which recently hit valuations of $500 billion and $183 billion, respectively.

Sticking to its core focus on the enterprise market, Cohere is positioning itself as a solution for companies prioritizing AI sovereignty—the need to retain data and model control within local jurisdictions. The latest $100 million extension included new backing from the Business Development Bank of Canada (BDC) and Nexxus Capital Management, the latter recognized for its investment network across Singapore.

Source: https://techcrunch.com/2025/09/24/cohere-hits-7b-valuation-a-month-after-its-last-raise-partners-with-amd/

Cohere Unveils North: A Privacy-First AI Agent Platform for Enterprises

August 6, 2025 - Canadian AI firm Cohere has launched North, a new AI agent platform designed to meet the strict data privacy and infrastructure needs of enterprises and government agencies. North allows full private deployment, ensuring that sensitive data remains within an organization’s control. It can operate on-premises, in hybrid or virtual private cloud environments, and even in air-gapped systems—requiring as few as two GPUs to run.

Enterprise-Grade Security Built In

North includes features like zero-trust architecture, fine-grained access controls, continuous red-teaming, third-party security assessments, and compliance with global standards such as GDPR, SOC-2, and ISO 27001. The platform ensures that data never leaves the customer’s infrastructure, alleviating concerns around external exposure or unauthorized model training.

Feature-Rich Out of the Box

At its core, North provides AI-powered chat and search capabilities. Users can handle tasks like answering internal or customer support questions, summarizing meetings, generating marketing content, and searching both internal databases and web sources. Responses include citation links and reasoning chains for transparency and auditability.

North is built on Cohere’s Command model family and its Compass search technology. According to Cohere, the version used in North has been fine-tuned specifically for enterprise-level reasoning, enabling it to generate documents, presentations, tables, and market research summaries.

Tool Integrations and Automation Support

The platform integrates with common enterprise tools such as Gmail, Outlook, Salesforce, Slack, and Linear. It also supports Model Context Protocol (MCP) to connect with proprietary or industry-specific applications, making it easy to embed within existing workflows. The system is designed to transition from assisting users to automating full tasks as confidence grows.

Strong Market Validation

North has already been piloted by several enterprise clients, including major players in finance, healthcare, tech, and defense. Cohere also offers specialized versions tailored to industry needs, such as “North for Banking,” which provides financial institutions with private LLM capabilities for regulated environments.

Strategic Momentum and Growth

This launch follows a period of rapid growth for Cohere, which recently reported a significant increase in annualized revenue, driven largely by private deployment demand. The company has also made strategic acquisitions—such as Ottogrid, a platform for automated market research—to enhance North’s capabilities further.

Source: https://techcrunch.com/2025/08/06/coheres-new-ai-agent-platform-north-promises-to-keep-enterprise-data-secure/

Cohere Unveils Aya Vision, a Best-in-Class Multimodal AI Model

March 4, 2025 - Cohere For AI, the nonprofit research arm of AI startup Cohere, has launched Aya Vision, an “open” multimodal AI model the lab describes as best-in-class.

Aya Vision is designed to handle tasks such as writing image captions, answering photo-related questions, translating text, and generating summaries in 23 major languages. Cohere is offering Aya Vision for free via WhatsApp, calling it “a significant step towards making technical breakthroughs accessible to researchers worldwide.”

“While AI has made significant progress, there is still a big gap in how well models perform across different languages — one that becomes even more noticeable in multimodal tasks that involve both text and images,” Cohere noted in a blog post. “Aya Vision aims to explicitly help close that gap.”

Aya Vision comes in two versions: Aya Vision 32B and Aya Vision 8B. The flagship Aya Vision 32B sets a “new frontier,” reportedly outperforming models twice its size, such as Meta’s Llama-3.2 90B Vision, on specific visual understanding benchmarks. The smaller Aya Vision 8B also delivers impressive results, scoring higher in some evaluations than models ten times its size.

Both models are available on Hugging Face under a Creative Commons 4.0 license with Cohere’s acceptable use addendum, restricting them from commercial use.

The models were trained on a “diverse pool” of English datasets, which Cohere translated and enriched with synthetic annotations — AI-generated tags that guide model training. This approach, increasingly adopted by other players like OpenAI, allowed Cohere to conserve resources while maintaining high performance. Gartner reports that synthetic data accounted for 60% of AI and analytics projects last year.

“This showcases our critical focus on efficiency and [doing] more using less compute,” Cohere wrote. “This also enables greater support for the research community, who often have more limited access to compute resources.”

Alongside Aya Vision, Cohere introduced AyaVisionBench, a benchmark suite evaluating a model’s abilities in “vision-language” tasks such as spotting differences between images and converting screenshots into code.

The AI industry faces an “evaluation crisis,” as many popular benchmarks fail to reflect real-world task proficiency. Cohere claims AyaVisionBench offers a “broad and challenging” framework to assess cross-lingual and multimodal understanding.

“[T]he dataset serves as a robust benchmark for evaluating vision-language models in multilingual and real-world settings,” Cohere researchers wrote on Hugging Face. “We make this evaluation set available to the research community to push forward multilingual multimodal evaluations.”

Source: https://techcrunch.com/2025/03/04/cohere-claims-its-new-aya-vision-ai-model-is-best-in-class/