Cohere Releases Command A+ — First Apache 2.0 Open Model with Lossless Quantization
Cohere unveils Command A+, a 218B-parameter sparse MoE model under Apache 2.0 license, featuring lossless 4-bit quantization that runs on a single Blackwell GPU.
Cohere Command A+ — What Makes It Special?
Cohere has released Command A+, a 218-billion-parameter language model under the highly permissive Apache 2.0 license. Despite its size, only 25 billion parameters are active during any generation step thanks to its Sparse Mixture-of-Experts architecture, making it remarkably efficient.
Why Is Lossless Quantization a Big Deal?
The standout feature is Cohere's W4A4 quantization — compressing the model to 4-bit precision with nearly zero quality loss. They achieved this by quantizing only the MoE experts while keeping attention pathways at full precision. The result: this 218B model runs on a single NVIDIA Blackwell B200 GPU or just two H100s.
Who Benefits from Apache 2.0 Licensing?
Unlike restrictive licenses from other frontier model providers, Apache 2.0 means anyone can use, modify, and commercialize Command A+ without strings attached. This is a major win for enterprises wanting sovereign AI — running frontier-grade models entirely within their own infrastructure.
What Are the Performance Numbers?
The W4A4 quantized version achieves 375 tokens per second with just 113ms time-to-first-token — a 63% speed increase and 17% latency reduction over the previous Command A Reasoning model. It excels at complex reasoning, multimodal document processing, and agentic workflows.
Frequently Asked Questions
Q: Can I run Command A+ locally? A: Yes, with the right hardware. The W4A4 version needs one Blackwell B200 or two H100 GPUs — significantly less than comparable models.
Q: What does Apache 2.0 allow? A: Commercial use, modification, distribution, and private use with no copyleft requirements. Attribution required, but no restrictions on how you use the model.
Q: How does it compare to GPT-5.5 or Claude Opus? A: While those are estimated to be in the trillions of parameters, Command A+ offers competitive reasoning quality with far lower compute requirements — especially valuable for self-hosted deployments.
Stay ahead of the AI curve. Follow @AiForSuccess for daily insights.
📬 Want more AI solopreneur insights?
Subscribe to our weekly newsletter →Related Articles
AI Model API Aggregation Platforms: From Simple Proxies to Enterprise AI Hubs
AI API aggregation platforms are evolving beyond protocol translation. Discover how these platforms are becoming essential infrastructure for enterprise AI adoption.
AI Jobs Explosion: 12x Increase in AI Positions Signals Massive Talent Demand
The AI job market has exploded with a 12x increase in AI positions since 2025. Discover what's driving this talent rush and what it means for your career in artificial intelligence.
Anthropic's Claude Code Source Leak: 1900 Files, 500K Lines of Code Gone Public
In June 2026, Anthropic accidentally published nearly 1,900 source files and 500,000 lines of Claude Code's core codebase. Here's what the leak revealed and why it matters for AI developers.