Cohere · Command R · Frontier
Command R+ pricing and limits
Cohere's retrieval-augmented generation model with built-in citations and tool use. Priced at the frontier tier, without a cache-read discount in the list price.
| Line | Tokens | Rate / 1M | Cost |
|---|---|---|---|
| Input, uncached | 10,000 | $2.50 | $0.0250 |
| Output | 800 | $10.00 | $0.00800 |
- Input / 1M
- $2.50
- Output / 1M
- $10.00
- Cached / 1M
- n/a
- Cache write / 1M
- n/a
- Blended 3:1
- $4.38
- Context
- 128K
What drives Command R+ cost
RAG workloads send retrieved documents with every request, so input dominates. Without a cache discount, controlling how many chunks you retrieve per query is the main cost lever.
Citation generation adds output tokens. For internal tools where citations are not shown, turning them off or using Command R for simple lookups lowers cost.
Cost of a typical call
Take a RAG answer with 10,000 input tokens and 800 output tokens. At Command R+ list prices ($2.50 input, $10.00 output per 1M) one call costs $0.0330, and 150,000 calls a month come to $4,950. The list price has no discounted cache-read rate, so every input token is billed at the full rate on every call.
Blended at a 3:1 input to output ratio, Command R+ costs $4.38 per 1M tokens. Routing the same call to the fallback in the policy below, Command R, costs $0.00198, which is what a tenant pays you in COGS once they cross the hard cap and PALM SOUQ downgrades instead of blocking.
Cost comparison
Command R+ against three alternatives
The same RAG answer priced on Command R, GPT-4.1, Mistral Large 2. Move the sliders to match your traffic; models without a cache-read rate bill every input token in full.
Benchmark quality per task before switching. A cheaper model that needs more steps or longer outputs can cost more per finished task than its rate suggests, which is why PALM SOUQ reports cost per feature, not only per token.
Command R+ this guide
$0.0330 per call
$3,300 / mo
baseline
Command R
$0.00198 per call
$198 / mo
-94% vs Command R+
GPT-4.1
$0.0264 per call
$2,640 / mo
-20% vs Command R+
Mistral Large 2
$0.0248 per call
$2,480 / mo
-25% vs Command R+
List prices from the PALM SOUQ LLM Price Index, computed per call as uncached input + cache reads + output. Reasoning tokens count as output on reasoning models.
Rate limits by usage tier
Cohere separates trial keys, which are rate limited and not for production, from production keys with higher limits per endpoint. Enterprise deployments on clouds follow the host's quotas.
Exact limits vary by account tier and change over time, so we do not reprint them here. Check the limits page for your own account in the Cohere documentation at docs.cohere.com. PALM SOUQ enforces your own per-tenant limits in front of the provider's, so a single customer cannot consume the shared allowance that every other tenant depends on.
Circuit-breaker best practices
Cap retrieved chunks per request at the application layer and enforce a per-request input ceiling at the gateway, so a retrieval bug that returns hundreds of documents is rejected before it is billed.
Use the tool thrashing check for multi-step tool use, where the model can repeat the same search with minor query changes.
When a breaker trips, the gateway answers in under 50 ms with HTTP 429 and X-PalmSouq-Reason: circuit_breaker_triggered, before the request reaches Cohere. Your agent framework sees a normal rate-limit response and stops, and the audit log records which heuristic fired and for which tenant.
# palmsouq.policy.yaml
policy: command-r-plus-business
match:
model: command-r-plus
tier: business # from X-PalmSouq-Tier
budgets:
monthly_usd: 300 # per tenant
soft_cap_pct: 80 # webhook to your app
on_hard_cap: fallback # or: block (HTTP 429)
fallback:
- command-r
breakers:
tool_thrashing: { identical_calls: 3, window_steps: 8 }
token_velocity: { max_tokens_per_s: 50000, window_s: 10 }
semantic_echo: { repeats: 3 }
max_steps: 10- budgetsPer-tenant monthly dollar cap with an 80% soft-cap webhook, keyed by X-PalmSouq-Tenant.
- fallbackAt the hard cap, requests reroute to Command R ($0.26 blended per 1M) instead of failing.
- breakersAll four loop heuristics on, evaluated in the request path with under 5 ms added latency.
Where it fits
Enterprise search and RAG with citations, especially where Cohere's private deployment options matter.
Provider documentation: docs.cohere.com
Related cost guides
Command R+ pricing questions
How much does Command R+ cost per million tokens?
At list price Command R+ costs $2.50 per 1M input tokens and $10.00 per 1M output tokens. Prices were last reviewed in the PALM SOUQ LLM Price Index and are for the standard tier and shortest context band.
Does Command R+ support prompt caching?
The Cohere list price for Command R+ has no discounted cache-read rate, so PALM SOUQ meters every input token at $2.50 per 1M. Shortening repeated context and caching full responses in your own application are the main savings levers.
What are the Command R+ rate limits?
Cohere separates trial keys, which are rate limited and not for production, from production keys with higher limits per endpoint. Enterprise deployments on clouds follow the host's quotas. For your account's exact numbers, see the Cohere documentation at docs.cohere.com.
How do I stop runaway Command R+ agent loops?
Cap retrieved chunks per request at the application layer and enforce a per-request input ceiling at the gateway, so a retrieval bug that returns hundreds of documents is rejected before it is billed. The sample policy on this page enables all four PALM SOUQ heuristics with a max step cap of 10 and falls back to Command R at the hard cap.
Is there a batch discount for Command R+?
No batch tier at list price. Use caching and model routing to lower cost for asynchronous jobs.
Govern Command R+ spend per customer.
Point your Cohere client at gateway.palmsouqltd.com, pass X-PalmSouq-Tenant, and every Command R+ token is attributed, capped and protected by circuit breakers.
14-day free production trial · 100k requests included · no card required