Skip to content

Every host for DeepSeek V3.2

×14 between cheapest and dearest input
HostBlendedInputOutputCachedPrecisionUptime
GMI Cloud
$0.234
$0.209$0.31$0.022fp899.9%
Atlas Cloud
$0.29
$0.26$0.38$0.13fp899.6%
DeepInfra
$0.29
$0.26$0.38$0.13fp499.7%
Venice
$0.299
$0.268$0.39$0.13—99.3%
Novita
$0.302
$0.269$0.40$0.135——
DeepSeek
$0.31
$0.28$0.40$0.028——
Gmi
$0.31
$0.28$0.40———
Baidu
$0.315
$0.28$0.42$0.028fp899.0%
SiliconFlow
$0.394
$0.259$0.80$0.14fp899.4%
Digitalocean
$0.465
$0.30$0.96$0.09—99.7%
Friendli
$0.75
$0.50$1.50$0.25—100.0%
Friendliai
$0.75
$0.50$1.50$0.25——
Google
$0.84
$0.56$1.68——0.0%
Azure
$0.855
$0.58$1.68———
Amazon Bedrock
$0.928
$0.62$1.85———
Phala
$1.00
$1.00$1.00$0.50—99.9%
Mara
$3.38
$3.00$4.50——0.0%
SambaNova
$3.38
$3.00$4.50——96.6%
Blended price: three input tokens for every output token, per 1M tokens.

Cheapest price over time

Per 1M tokens
Cheapest input and output price for DeepSeek V3.2 over timeFree$0.10$0.20$0.30$0.402 Dec13 May23 Oct$0.31$0.209
  • Input
  • Output
The cheapest list price at any host on each day it changed.

What your workload costs

A month, at every host
Cheapest a month
$93.60
GMI Cloud
  • GMI Cloud
    $93.60
  • Atlas Cloud
    $116.00
  • DeepInfra
    $116.00
  • Venice
    $119.51
  • Novita
    $120.70
  • DeepSeek
    $124.00
  • Gmi
    $124.00
  • Baidu
    $126.00
  • SiliconFlow
    $157.70
  • Digitalocean
    $186.00
  • Friendli
    $300.00
  • Friendliai
    $300.00

6 more hosts cost more.

Rent GPUs or pay per token?

Break-even
Renting the GPUs
$16,878/mo
Break-even volume
72.1B tokens/mo
Needed around the clock
27.4K tok/s

8 × H100 cost $16,878 a month. The cheapest API (GMI Cloud) charges $0.234 per 1M blended tokens, so the GPUs only pay off above 72.1B tokens a month: the box must serve 27.4K tokens a second, every second.

GPU price: cheapest on-demand offer per GPU-hour × 730 hours. No throughput is assumed: it depends on the model size, precision, batch size and serving software, so measure yours. Power, storage, egress and engineering time are not included.

LMArena rating over time

Text leaderboard
LMArena text rating of DeepSeek V3.2 over time1,4001,4101,4201,4303 Oct12 Apr20 Oct1,425

Rated about the same, by price

Blended, per 1M tokens

Quality data: LMArena (CC BY 4.0), Epoch AI (CC BY 4.0). Prices: OpenRouter, LiteLLM (MIT), provider pages. List prices in US dollars before discounts and taxes; other currencies converted at the European Central Bank reference rate. The figures on these pages are free to reuse under CC BY 4.0 with a link to this page.