DeepSeek V4 Pro API pricing

6 providers serve DeepSeek V4 Pro. DeepSeek V4 Pro is an open-weight model (MIT) with 1600B total parameters, 49B active per token (mixture-of-experts), up to 1049K context. Prices below are per 1M tokens, cheapest input first (self-hosting math further down).

Refreshed 2026-08-23 · list prices from provider APIs

Input floor $0.435/1M at Tencent Output floor $0.870/1M at Tencent Find your workload winner →

Provider	Input $/1M	Output $/1M	Cache read $/1M	Context
Tencent	$0.435	$0.870output floor	$0.0036	1000K
OpenRouter	$1.32	$3.96	-	1049K	Use
DeepSeek	$1.32	$3.96	$0.044	1000K	Use
Azure AI Foundry	$1.74	$3.48	-	1000K	Use
Fireworks AI	$1.74	$3.48	$0.145	1049K	Use
Alibaba Model Studio	$2.40	$4.80	$0.200	1000K	Use

The cheapest way to run DeepSeek V4 Pro

On an illustrative 70M-input / 30M-output monthly workload, today's lowest listed cost is roughly $56.55/month, at Tencent. Input-heavy, output-heavy and cache-heavy workloads can produce different winners. Use the live mix calculator below rather than combining floors from two different providers. On output-token cost alone, the API floor remains below our self-hosting estimate of ~$4.04/1M. Details below. Run your own numbers in the breakeven calculator, or read the breakeven math.

Your actual monthly cost

Uses today's list prices

Workload preset Input tokens / month (millions) Output tokens / month (millions) Input served from cache %

Cache read uses the listed cache rate where available; otherwise it falls back to normal input price. Batch, write-cache, volume and negotiated discounts are excluded.

Or run it yourself: DeepSeek V4 Pro self-hosting economics

DeepSeek V4 Pro is open-weight (MIT), so the API price above competes with the GPU-hour market. 1600B total parameters (MoE, ~49B active per token) needs roughly 2201 GB VRAM at FP8 or 1101 GB at INT4, KV-cache headroom included.

No single GPU fits DeepSeek V4 Pro at FP8 (2201 GB needed), so it needs a multi-GPU node. Cheapest tracked option: 12× MI300X (192 GB each) at roughly $6.00/hr total (RunPod). Rent 12× MI300X at RunPod → INT4 quantization drops the requirement to 1101 GB. Tune it in the calculator.

Breakeven estimate: a well-batched vLLM deployment on 12× MI300X ($6.00/hr) at ~826 aggregate tok/s and 50% utilization works out to roughly $4.04 per 1M output tokens, versus $0.870 via the cheapest output-token API (Tencent). The API is cheaper unless your utilization or throughput beats these assumptions. Planning estimate: throughput varies with hardware, quantization, batch size and context. Tune it in the calculator.

Related models

Model	Cheapest in $/1M	Cheapest out $/1M	Providers
DeepSeek R1	$0.200	$0.400	17
DeepSeek V3	$0.200	$0.200	16
DeepSeek R1 Distill Llama 70B	$0.200	$0.375	10
DeepSeek V3.2	$0.269	$0.400	9
DeepSeek-V3.1	$0.270	$1.00	9
DeepSeek V4 Flash	$0.050	$0.101	9

FAQ

What is the cheapest API for DeepSeek V4 Pro?

As of 2026-08-23, the lowest input price for DeepSeek V4 Pro is Tencent at $0.435 per 1M input tokens. The lowest output price is Tencent at $0.870 per 1M output tokens. The cheapest provider for a real workload depends on its input/output mix. The next-lowest input price is OpenRouter at $1.32, a 203% difference.

How much VRAM do you need to self-host DeepSeek V4 Pro?

DeepSeek V4 Pro has 1600B parameters, so plan for roughly 2201 GB of VRAM at FP8 or 1101 GB at INT4/AWQ, KV-cache headroom included. That exceeds a single GPU. A typical node is 12× MI300X.

← All models · GPU rental prices · Breakeven calculator · Get this data as JSON