Skip to main content
9 – 17 UHR +49 8031 3508270 LUITPOLDSTR. 9, 83022 ROSENHEIM
DE / EN
LLM Cohere Canada

Cohere Command

Cohere Command A+ (MoE, 218B/25B active, Apache 2.0, May 2026), Command A, Command A Reasoning, Vision and Translate – enterprise LLMs with strong RAG focus. Merger with Aleph Alpha (April 2026). AI consulting from Germany for Cohere integration.

License Proprietary (Command), CC-BY-NC (Aya)
GDPR Hosting Available
Context 256k (Command A), 128k (Command R+) Tokens
Modality Text, Image → Text

Versions

Overview of available model variants

ModelReleaseEUStrengthsWeaknessesStatus
Command A+ (command-a-plus-05-2026) Recommended
May 2026
New flagship: Sparse MoE with 218B total parameters (25B active) Unifies Command A, Reasoning, Vision and Translate into a single model Apache 2.0 license – fully open weights 48 languages including all official EU languages Native citations + W4A4 lossless quantization 150% higher throughput than Command R+ 08-2024
Freshly released – ecosystem still maturing
Current
Command A (command-a-03-2025)
March 2025
Proven flagship (111B parameters, 256k context) Strong tool use and agentic AI Open weights
Premium pricing ($2.50/$10.00 per 1M tokens) Superseded by Command A+ (May 2026)
Current
Command A Reasoning
August 2025
Cohere's first hybrid reasoning model 256k context, 111B parameters Strongest tool use in the Command family
Higher latency when reasoning is enabled
Current
Command A Vision
2025
First multimodal Command model Image understanding, document OCR
Current
Command A Translate
2026
Cohere's first enterprise translation model 23 languages, state-of-the-art quality
Current
Command R7B
December 2024
Compact 7B model Good fit for edge / self-hosting
Lower capacity than Command A
Current
Aya Vision
March 2025
Multilingual image understanding Image description and translation
CC-BY-NC licence (research only)
Current
Command R+ (command-r-plus-08-2024)
2024-08
104B parameters – powerful model 50% higher throughput than predecessor 25% lower latency Strong RAG capabilities with reliable citations Improved structured outputs (JSON) 128k context window
Weaker in coding than specialised models
Current
Command R
2024
Best price-performance ratio RAG-optimised 128k context window
Current
Command Light
2024
Cost-efficient Fast
Lower capacity
Current
Aya Expanse
2024
Multilingual (23 languages) Open weights
Current

Use Cases

Typical applications for this model

RAG Applications
Enterprise Search
Document Analysis
Knowledge Management
Customer Service
Agentic Workflows
Multilingual Applications

Technical Details

API, features and capabilities

API & Availability
Availability Public
Requests/Min 10000
Tokens/Min 100000
Latency (TTFT) ~400ms
Throughput ~150 Tokens/Sec
Features & Capabilities
Tool Use Function Calling Structured Output Vision Reasoning Mode Web Browsing File Upload
Training & Knowledge
Knowledge Cutoff Early 2025 (Command A)
Fine-Tuning Available (Fine-tuning API, Custom Models)
Language Support
Best Quality English, German, French, Spanish
Supported 23 languages (Aya)
Particularly strong in English, good quality in German

Hosting & Compliance

GDPR-compliant hosting options and licensing

GDPR-Compliant Hosting Options
AWS
Frankfurt (eu-central-1)
Amazon Bedrock
Azure
West Europe, Germany West Central
Azure AI Foundry
Google Cloud
Frankfurt (europe-west3)
Vertex AI
License & Hosting
License Proprietary (Command), CC-BY-NC (Aya)
Security Filters Customizable
Enterprise Support Yes
SLA Available Yes
On-Premise

Benchmarks

Performance comparison with standardized tests

MMLU
75
HumanEval
62

As AI consultants based in Rosenheim, Germany, we recommend Cohere Command for enterprises in the DACH region (Germany, Austria, Switzerland) focusing on RAG, Enterprise Search, and Agentic AI. With the new flagship Command A+ (May 2026, MoE with 218B/25B active parameters, Apache 2.0), the proven Command A and Command R+, Cohere offers powerful solutions for knowledge-based applications – GDPR-compliant via AWS Bedrock Frankfurt, Azure AI Foundry, Google Vertex AI and Microsoft Foundry. Full on-premise deployment is also available.

Update May 2026: On 20 May 2026, Cohere released Command A+ (command-a-plus-05-2026) – the Command family’s first fully Apache 2.0-licensed frontier model. The Sparse MoE model (218B total, 25B active parameters) unifies vision, reasoning, translation and agentic workflows in a single model, supports 48 languages including all official EU languages, and runs on as little as 1x B200 or 2x H100 thanks to W4A4 quantization.

Update April 2026: On 24 April 2026, Cohere announced the acquisition of and merger with Aleph Alpha (Heidelberg) – combined valuation approximately USD 20bn, with a USD 600m strategic investment from the Schwarz Group. This creates a transatlantic sovereign-AI platform with a German footprint. See Aleph Alpha for details.

Command A+ – The New Flagship (May 2026)

With Command A+ (command-a-plus-05-2026), Cohere released the Command family’s first Mixture-of-Experts model on 20 May 2026 – and the first frontier model under Apache 2.0:

  • Sparse MoE: 218B total parameters, 25B active parameters per token
  • Unified capabilities: vision, reasoning, translation and agentic tool use in a single model
  • Efficiency: runs on 1x B200 or 2x H100 thanks to W4A4 lossless quantization, 150% higher throughput than Command R+ 08-2024
  • 48 languages including all official EU languages
  • Native citations with span-level accuracy out of the box
  • 128k context window
  • Apache 2.0 license – maximum control for sovereign-AI deployments

Command A+ is positioned for sovereign critical infrastructure and is available via the Cohere API, Microsoft Foundry and Hugging Face.

Command A – The Previous Flagship

Since March 2025, Command A (command-a-03-2025) has established itself as a reliable foundation for enterprise applications – and remains a production-stable option even after the release of Command A+:

  • Strongest Tool Use and Agentic AI: Multi-step reasoning and autonomous task execution for complex workflows
  • State-of-the-art Translation: Translation across 23 languages at the highest level
  • General Use: Optimised for versatile enterprise applications

Command A Translation is a dedicated variant specifically developed for multilingual workflows – ideal for internationally operating DACH companies.

New Models: Reasoning, Vision and Aya Vision

Command A Reasoning

For tasks requiring deep logical thinking, Cohere has offered Command A Reasoning since 2025 – with transparent chain-of-thought processing.

Command A Vision

Command A Vision extends the Command family with image understanding: document OCR, image analysis and multimodal RAG workflows are now possible.

Aya Vision

The open-weights model Aya Vision (March 2025) by Cohere Labs enables multilingual image understanding – ideal for research (CC-BY-NC licence).

Cohere North

Cohere North is the enterprise platform for secure AI workspaces. Available since January 2025, North is used by companies such as RBC, Dell and Bell Canada.

Command R+ – Proven RAG Performance

Command R+ (command-r-plus-08-2024) remains a strong choice for RAG-intensive applications with its 104B parameters:

  • 50% higher throughput – faster processing of large data volumes
  • 25% lower latency – more responsive applications
  • Reliable citations – native source references with span-level accuracy
  • Improved structured outputs – reliable JSON generation
  • 128k context window – processing of extensive documents (~320 pages)

RAG & Enterprise Search – Cohere’s DNA

Cohere differs fundamentally from OpenAI and Anthropic: RAG is not an add-on feature, but the core competency. Native citation support with span-level accuracy is unique in the market. The combination of Command models, Embed v3 (leading embedding model), and Rerank enables:

  • Grounded Generation – answers based on your documents
  • Native Citations – every statement with precise source reference at paragraph level
  • Precise Reranking – optimised relevance sorting
  • 128k Context Window – processing of extensive documents (~320 pages)

EU Availability & GDPR Compliance

Cohere Command is available GDPR-compliant in the EU – with multiple hosting options:

  • AWS Bedrock: Frankfurt (eu-central-1)
  • Azure AI Foundry: West Europe, Germany West Central
  • Google Vertex AI: Frankfurt (europe-west3)
  • On-Premise: Full deployment for maximum data sovereignty

Broad availability across all major cloud providers combined with the on-premise option makes Cohere a flexible choice for DACH enterprises with strict data protection requirements.

Integration with CompanyGPT

Cohere Command is fully integrated into CompanyGPT. Use Cohere’s RAG strengths to make your internal knowledge searchable – GDPR-compliant in German or European data centres or fully on-premise.

Our Recommendation

Command A+ (May 2026) is now the first choice for Agentic Workflows, Tool Use, multimodal and multilingual applications – especially for organisations requiring Apache 2.0 licensing, sovereign deployment and EU language coverage. Command A remains the robust choice for workloads already in production. For classic RAG applications with high throughput, we recommend Command R+ with 104B parameters – the best price-performance ratio for Enterprise Search. For cost-efficient scenarios with less complex requirements, Command R or Command Light are well suited.

Consider weaknesses: Cohere is weaker in code generation than specialised models. For coding use cases, we recommend OpenAI GPT or Anthropic Claude.

Merger Talks with Aleph Alpha (April 2026)

Cohere and Aleph Alpha (Heidelberg) are reportedly in talks about a potential merger, supported by the German federal government. Such a merger could combine Cohere’s RAG expertise with Aleph Alpha’s sovereignty focus for the European market.

Cost estimation for this model

For up-to-date token pricing, model variants and EU availability, see our sister project ai-prices.eu. It helps you compare and estimate the operational cost of leading AI models for your specific use case.

Compare prices on ai-prices.eu

ai-prices.eu is a project by innFactory AI Consulting GmbH and provides transparent cost estimates for leading AI models.

Consultation for this model?

We help you select and integrate the right AI model for your use case.