DeepSeek V3.2 API pricing

What does DeepSeek V3.2 cost per request?

A typical request of 5,000 input tokens and 1,500 output tokens costs about $0.002 at list price. A thousand of those requests costs about $2.03.

Model usage is drawn from your balance at the upstream list price of each model. Figures above are indicative and exclude any plan fee; the console is the source of truth for live pricing.

Where DeepSeek V3.2 fits

DeepSeek V3.2 has the cheapest output price on the gateway at $0.42 per 1M tokens, which changes which workloads are economically viable in the first place.

Pick it when

• You are generating a lot of text rather than reading it, so the output price dominates the bill.
• You are running bulk jobs - enrichment, drafting, translation - where a per-call cost above a fraction of a cent kills the idea.
• You want a cheap default for internal tools that have no revenue attached to each call.

What to watch

• The context window is 128K, which is smaller than the 400K and 1M windows elsewhere in the catalogue.
• Frontier-level reasoning is better served by a stronger model, so pair it with an escalation path for the hard 5%.

Best for: Bulk generation, enrichment pipelines and internal tooling where unit cost sets the ceiling.

How to call DeepSeek V3.2 through the gateway

Point your client at https://api.quickrouter.homes/v1 with a QuickRouter key and pass deepseek-v3.2 as the model name. Nothing else in an OpenAI-style client changes.

curl https://api.quickrouter.homes/v1/chat/completions \
  -H "Authorization: Bearer sk-qr-your-key" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "deepseek-v3.2",
    "messages": [{"role": "user", "content": "Hello"}]
  }'

The console lists the model ids your key can reach - use the id exactly as it appears there.