The latest AI news and updates of Cohere — AI company building enterprise models for search, RAG, and business applications. Covering Cohere's latest company news, launches, and product updates from the past 90 days.
Cohere15h agoCohere highlights rapid community adoption of North Mini Code two days after its release. Developers have already delivered llama.cpp support via PR #24260, Unsloth GGUF quantizations ranging from 9GB to 61GB, and day-zero MLX integration. The community also built Taskflow, a CLI task manager, and upgraded documentation to support autonomous coding workflows.
Cohere16h agoCohere Transcribe ranked first on the new Hugging Face Far-Field ASR benchmark with a 17.9% word error rate. The model outperformed IBM Granite Speech by nearly 2 points and NVIDIA Parakeet by 3.6 points. This benchmark measures transcription accuracy in real-world, far-field audio environments like meeting rooms and contact centers, where the model ranked first across every metric.
Cohere16h agoCohere Labs published a report identifying a widening gap between static 2023 AI labor exposure scores and current policy needs. The research argues that relying on outdated, task-based metrics to inform 2026 labor policy compounds limitations. It calls for policymakers and researchers to adopt dynamic, worker-centered measurement tools that reflect actual AI capabilities and real-world labor market outcomes.
CohereJun 10Cohere introduced North Mini Code, its first open-source coding model. This small, efficient model is designed for agentic performance, offering developers control over their coding infrastructure for autonomous workflows.
CohereJun 4Cohere secured first place in NATO's Agentic AI for Cognitive Warfare Innovation Challenge. This recognition highlights the growing role of agentic AI in helping democratic nations address information threats.
CohereMay 29Cohere released new benchmarks showing its Command A+ model outperforms specialist systems like Google Translate and frontier models in machine translation across 48 languages. A new tokenizer also reduces inference costs by up to 20% for non-Latin languages, making sovereign multilingual AI more practical for regulated industries.
CohereMay 21Cohere released W4A4 quantized weights for its 218-billion parameter Command A+ model, enabling frontier-class reasoning on a single NVIDIA B200 GPU. By using quantization-aware distillation to maintain performance, the update allows enterprises to deploy massive agentic models with a significantly smaller hardware footprint.
CohereMay 20Cohere released Command A+, a 218-billion parameter open-source model optimized for complex reasoning and multimodal agentic tasks. By achieving high performance on as little as two H100 GPUs, the model allows enterprises to deploy frontier-class agents entirely within their own private infrastructure.
CohereMay 20Cohere signed strategic agreements with Indra Group and Multiverse Computing to deploy sovereign AI models for government and defense sectors. The partnerships focus on local data control and regional language support, moving beyond the cloud-rental model toward full organizational ownership of AI infrastructure.
CohereMay 19Cohere acquired Reliant AI to accelerate the development of North for Pharma, a specialized agentic AI platform for biopharma research and clinical analytics. The move deepens Cohere's vertical strategy by providing secure, sovereign AI systems for highly regulated life sciences organizations.
CohereApr 24Cohere and Aleph Alpha formed a transatlantic partnership to build sovereign, enterprise-grade AI anchored in Canada and Germany. The alliance combines global scale with European research to offer businesses and governments a secure alternative to US-based AI ecosystems while maintaining strict data residency.
CohereApr 24Cohere validated that Mixture-of-Experts models achieve higher speedups from speculative decoding than dense models by staying in a memory-bandwidth-bound sweet spot. The research shows that consecutive tokens naturally reuse the same experts, significantly reducing the data-loading bottleneck during parallel verification.
CohereApr 24Cohere released production-ready W4A8 quantization kernels for dense and Mixture of Experts models, now integrated into the vLLM inference framework. By combining 4-bit weights with 8-bit activations, the update achieves up to 58 percent faster prefill and 45 percent faster decoding on NVIDIA Hopper GPUs.
CohereApr 13Cohere partnered with RWS Group to fine-tune a 100B+ parameter Command model that now powers the Language Weaver Pro translation platform. This specialized model outperforms general-purpose frontier models by combining massive scale with deep linguistic and cultural expertise.
CohereApr 1Cohere and Ensemble Health Partners are developing a custom large language model specifically for healthcare revenue cycle management. By training on proprietary operational logs and expert annotations, the model handles complex financial workflows that general-purpose AI wrappers struggle to navigate.
CohereMar 28Cohere launched Transcribe, a 2-billion parameter open-source speech recognition model that currently holds the top spot on the HuggingFace Open ASR Leaderboard. By achieving a 5.42% word error rate, it provides a high-accuracy, high-throughput alternative for enterprise workflows that previously relied on larger or proprietary models.
CohereMar 18Cohere announced NVIDIA ecosystem-native models and an optimized version of its agentic North platform for NVIDIA DGX Spark. The partnership targets regulated industries that need high-performance AI running securely on-premise within national borders.
Frequently asked questions