Cohere API Pricing: Every Model Compared (April 2026)

Trial (Free)

• ✓ 100 API calls per minute
• ✓ 1,000 calls per month
• ✓ Access to all models
• ✓ No credit card required
• ✓ Enough for prototyping and evaluation

Command R+

• ✓ Most capable text generation model
• ✓ Strong at RAG and tool use
• ✓ 128K context window
• ✓ Multilingual support (10+ languages)
• ✓ Cheaper than Sonnet 4.6 and GPT-4.1

Command R

• ✓ Budget text generation model
• ✓ Good for classification and extraction
• ✓ 128K context window
• ✓ Comparable to GPT-4.1 Nano on simple tasks
• ✓ very cost-effective at scale

Embed v3

• ✓ High-quality text embeddings
• ✓ 1024 dimensions (configurable)
• ✓ Multilingual support
• ✓ Comparable to OpenAI text-embedding-3-small
• ✓ Top for search and RAG

Rerank 3.5

• ✓ Reranks up to 100 docs per search
• ✓ Dramatically improves RAG accuracy
• ✓ Works with any vector database
• ✓ Simple API, just send query + documents
• ✓ Documents >500 tokens split into chunks

• ⚠ Rerank charges per search, not per document. One search with 100 documents costs the same as one search with 10 documents (if all docs are under 500 tokens). But documents over 500 tokens get split into chunks, and each chunk counts separately.
• ⚠ Command R+ output tokens cost 4x input tokens. Long generative responses get expensive. Use Command R ($0.15/$0.60) for tasks that don't need R+ quality.
• ⚠ The free trial's 1,000 calls/month is generous for prototyping but not for production. There's no pay-as-you-go middle ground, you go from free to production pricing.
• ⚠ Embed v3 at $0.10/1M tokens is cheap, but embedding large document collections adds up. A million 500-word documents is roughly 650M tokens, costing $65 to embed.
• ⚠ Cohere's pricing is competitive with OpenAI and Anthropic, but the model quality gap matters. Command R+ is strong at RAG but may lag behind Sonnet 4.6 or GPT-4.1 on general reasoning and coding tasks.

RAG pipeline builder

Embed v3 ($0.10/1M tokens) for embeddings + Rerank 3.5 ($2/1K searches) for result quality. This combo is Cohere's strongest use case and competitive with any alternative.

Text generation at scale

Command R ($0.15/$0.60) for high-volume simple tasks. Command R+ ($2.50/$10) when you need stronger reasoning. Both are cheaper than Anthropic and OpenAI equivalents.

Enterprise with compliance needs

Cohere offers deployment on your own cloud (VPC) and on-premises options. Contact sales for enterprise pricing. This is a differentiator vs OpenAI and Anthropic, which are API-only.