DeepSeek V4 Pro API pricing on Deep Infra
Every Deep Infra DeepSeek V4 Pro rate we track, compared against 22 providers serving the same model.
The cheapest DeepSeek V4 Pro input price we currently track is $0.600/1M on IteraCompute.
The cheapest DeepSeek V4 Pro output price is $1.79/1M on IteraCompute.
Deep Infra DeepSeek V4 Pro rates
Prices per 1M tokens. Batch and cached-input rates are shown where the provider publishes them.
| Mode | Input | Output |
|---|---|---|
| Standard | $1.30/1M | $2.60/1M |
About DeepSeek V4 Pro
- Creator
- DeepSeek
- Context
- 1.0M
- Modalities
- text
- Open weights
- Yes
DeepSeek V4 Pro on other providers
- IteraCompute$0.600/1M
- EcoHash$0.910/1M
- OpenRouter$0.955/1M
- GMI Cloud$0.957/1M
- IO.NET$1.21/1M
- RouterPlus$1.30/1M
- Baseten$1.32/1M
- DeepSeek$1.32/1M
- DigitalOcean$1.32/1M
- Empirio Labs$1.32/1M
- Novita AI$1.32/1M
- Perplexity$1.32/1M
- PopNow$1.32/1M
- Together AI$1.32/1M
- Runcrate$1.36/1M
- Geodd$1.48/1M
- HiAPI$1.50/1M
- Crusoe$1.74/1M
- Prime Intellect$1.91/1M
- Lyceum$2.00/1M
- Heabsy$2.20/1M
About Deep Infra
Deep Infra is an AI inference cloud that provides serverless APIs for 100+ models and dedicated GPU rentals. Features OpenAI-compatible endpoints, custom model deployments, and infrastructure optimized for scale.
Frequently Asked Questions
How much does DeepSeek V4 Pro cost on Deep Infra?
Deep Infra serves DeepSeek V4 Pro from $1.30 per 1M input tokens, across 1 tracked pricing mode. Prices are collected daily — see the table above for current input, output, and cached-input rates.
Is Deep Infra the cheapest way to run DeepSeek V4 Pro?
Deep Infra ranks #6 of 22 providers we track serving DeepSeek V4 Pro on input price. The cheapest input price right now is on IteraCompute. The lowest output price is $1.79/1M on IteraCompute. Rates change often, and throughput, latency and context limits differ between providers — check the comparison above before committing.
Does Deep Infra offer batch or cached pricing for DeepSeek V4 Pro?
We track 1 DeepSeek V4 Pro offering from Deep Infra. Batch and cached-input rates appear in the table above when the provider publishes them; blank cells mean we have no current figure for that mode.