API Pricing
Cohere API Pricing 2026:
Command R+, Embeddings & Rerank
Cohere is the enterprise-focused LLM provider built for RAG, search, and business applications. Complete 2026 pricing for Command R+, Command R, embeddings, and reranking.
9 min read·Updated March 2026
Cohere Pricing Summary
$2.50
Command R+ input per 1M
$0.10
Command R input per 1M
$0.10
Embed English per 1M tokens
FREE
developer trial tier
Cohere API Pricing 2026 — All Models
| Model | Input (per 1M) | Output (per 1M) | Best For |
|---|---|---|---|
| Command R+ | $2.50 | $10.00 | Complex RAG, enterprise Q&A |
| Command R | $0.15 | $0.60 | Cost-effective RAG, summarization |
| Command R+ 08-2024 | $2.50 | $10.00 | Multilingual (10 languages) |
| Embed English v3 | $0.10 | N/A | Semantic search, clustering |
| Embed Multilingual v3 | $0.10 | N/A | 100+ languages |
| Rerank v3.5 | $2.00 | per 1K docs | RAG result reranking |
Where Cohere Excels: RAG & Enterprise Search
Cohere's unique offering vs OpenAI/Anthropic is their focus on the full enterprise search stack:
- Command R is specifically trained for RAG use cases — handles 128K context, retrieves from tool-call citations
- Embed models create document embeddings optimized for enterprise search quality
- Rerank model re-orders retrieved chunks before passing to the LLM — significantly improves RAG quality
Cohere vs OpenAI for RAG Applications
| Component | Cohere Stack | Cost | OpenAI Stack | Cost |
|---|---|---|---|---|
| Embeddings | Embed English v3 | $0.10/M | text-embedding-3-small | $0.02/M |
| Reranking | Rerank v3.5 | $2.00/1K | No native option | N/A |
| Generation | Command R | $0.15/M in | GPT-4o mini | $0.15/M in |
For RAG: Cohere's reranking step costs $2/1K docs but typically reduces the number of chunks needed in the context window, often saving more than it costs on generation tokens.
Cohere RAG Cost Example
Internal knowledge base (50K queries/month)
- Embed 1M documents (one-time): 1B tokens × $0.10/M = $100 one-time
- Per query: embed query (50 tokens), retrieve 20 chunks, rerank, generate answer
- Rerank cost: 50K × 20 docs / 1K = 1,000 rerank calls × $2 = $2,000/month
- Generation (Command R, 3K tokens/query): 150M tokens × $0.375/M avg = $56/month
- Total: ~$2,056/month
Note: For RAG at scale, the reranking cost often dominates. Consider reranking only top 10 (not 20) chunks, or only applying reranking when confidence is low.
Cohere Free Tier
Cohere offers a generous developer trial:
- 5 requests/minute on all models (free)
- No credit card required for trial
- Full access to Command R, embeddings, reranking
- Production tier starts at pay-as-you-go with no minimum spend
Calculate Your RAG Stack Costs
Compare Cohere, OpenAI, and Gemini for your search application.
AI Cost Calculator