Cohere Releases Command A+ Open-Source: Advancing Sovereign AI for Developers
Key Takeaways
- ▸Command A+ is released under Apache 2.0 license with weights available on Hugging Face and multiple quantizations for flexible deployment
- ▸Outperforms previous Command models significantly on agentic tasks (25% on Terminal-Bench Hard vs. 3% previously) and reasoning benchmarks
- ▸Designed for sovereign AI—developers can run, control, and optimize the model in their own environments without platform dependency
Summary
Cohere has announced the open-source release of Command A+, a mixture-of-experts (MoE) language model optimized for high-performance agentic tasks. The model, available under an Apache 2.0 license on Hugging Face, represents a consolidation of the Command series capabilities and is designed to advance Cohere's mission of sovereign AI—giving developers direct access to enterprise-grade models they can run, control, and adapt locally.
Developed through a year of deploying North, Cohere's enterprise workspace for building and deploying agentic AI, Command A+ significantly outperforms previous generation Command models across multiple benchmarks. It shows substantial improvements in agentic reasoning (τ²-Bench Telecom scores improved from 37% to 85%), multimodal understanding, and retrieval tasks, while also expanding multilingual support from 23 to 48 languages. The model scores 37 on the Artificial Analysis Intelligence Index, outperforming other leading open-source models.
The release emphasizes efficiency and practical deployment: Command A+ supports low-bit quantization, efficient inference, and integration with open inference frameworks. Developers can deploy the model locally or use Cohere's Model Vault for managed inference, making it accessible for both experimentation and production workflows.
- Supports efficient inference with low-bit quantization and expands multilingual capabilities to 48 languages for enterprise workloads
Editorial Opinion
The release of Command A+ as open-source represents a strategic commitment to democratizing enterprise-grade AI capabilities. By prioritizing sovereignty and local deployment, Cohere is directly addressing the industry's growing concern about vendor lock-in and model dependency. This approach—combined with strong agentic performance and efficiency gains—positions open-source models as genuinely competitive alternatives to closed commercial offerings. If community adoption matches the technical quality, this release could meaningfully shift the balance toward decentralized AI infrastructure.
