Cohere Command A+ AI features and pricing (October 2026)

Command A+ is Cohere's current flagship language model, a sparse mixture-of-experts model with reasoning, vision and translation in 48 languages, available through Cohere's deployments and as open weights under Apache 2.0.

Command A+ launched on May 20, 2026 as Cohere's first mixture-of-experts model, with 218 billion total and 25 billion active parameters and a 128,000-token context. It takes text and images, produces text and supports configurable reasoning. The weights are on Hugging Face under Apache 2.0 in bf16, fp8 and w4a4 versions; the w4a4 version runs on one B200 or two H100 GPUs, while bf16 needs four B200 or eight H100. Cohere also offers Command A Reasoning, Vision and Translate as separate models.

• Works across European languagesSupport for all official EU languages suits multinational document and customer workflows. Teams avoid separate models per language.
• Runs reasoning on private infrastructureOpen weights and a modest hardware footprint let companies run a reasoning model in their own data center. Regulated industries keep data in house.
• Reads documents with imagesVision input covers charts and scanned pages. Finance and legal teams extract information from mixed documents.

• Apache 2.0 open weights for a model with reasoning and vision.
• Canadian provider with deployment in VPCs, on premises or on major clouds.

Cohere was founded in Toronto in 2019 and is headquartered there.

Ask Cohere for the current API price, since it was not on the pricing page at our check. Self-hosting needs at least one B200 or two H100 GPUs for the w4a4 version, more for fp8 and bf16, and your own operations team. Trial keys are free but rate limited and may not be used for production or commercial purposes.

No reviews yet. Be the first to share how your team uses it.

Every review shows its verification level. We read each one before it goes live. How we check reviews