Home / DeepSeek / DeepSeek V4 Pro (latest)
DeepSeek V4 Pro (latest) API price
deepseek/deepseek-v4-pro
The cheapest DeepSeek V4 Pro (latest) API is $0.35/M input at Nahcrof (inference host) - all 22 providers compared below, per 1M tokens.
Rolling name - always points to the newest snapshot of this model. The pinned snapshots are priced separately: DeepSeek V4 Pro (0813 snapshot).
FlagshipDeepSeek1.0M context22 sellers
Input
$0.35
per 1M tokens · Nahcrof (inference host)
up to $1.91
Output
$0.80
per 1M tokens · Nahcrof (inference host)
up to $3.96
Where to buy it
Cheapest DeepSeek V4 Pro (latest) API providers compared - 22 sellers, from $0.35/M.
| Provider | Input $/M | Output $/M | Cached in | TPS | Uptime |
|---|---|---|---|---|---|
| Nahcrof (inference host)Q8_0 | 0.350cheapest | 0.800 | 0.0030 | 11 | - |
| OrcaRouter (aggregator) | 0.442 | 0.884 | - | - | - |
| StreamLake (via OpenRouter)fp8 | 0.870 | 1.74 | 0.072 | - | 100% |
| GMICloud (via OpenRouter)fp8 | 0.792 | 2.38 | 0.026 | - | 97.6% |
| Ionstream (via OpenRouter)fp4 | 1.13 | 2.26 | 0.094 | - | 99.4% |
| CoreWeave (via OpenRouter)fp8 | 1.15 | 2.55 | 0.200 | - | 100% |
| DeepInfrafp8 | 1.30 | 2.60 | 0.100 | - | 97.7% |
| Alibaba (via OpenRouter)fp8 | 1.42 | 2.83 | 0.118 | - | 100% |
| SiliconFlow (via OpenRouter)fp8 | 1.50 | 3.13 | 0.135 | - | 100% |
| Novitafp8 | 1.60 | 3.20 | 0.135 | - | 100% |
| Venice | 1.65 | 3.30 | 0.330 | - | 47.4% |
| AtlasCloudfp4 | 1.68 | 3.38 | 0.130 | - | 99.1% |
| Baidu (via OpenRouter)fp8 | 1.69 | 3.38 | 0.140 | - | 100% |
| NextBit (via OpenRouter)fp8 | 1.72 | 3.45 | 0.140 | - | 93.3% |
| BaseTen (via OpenRouter)fp4 | 1.74 | 3.48 | 0.145 | - | 99.6% |
| DigitalOcean (via OpenRouter) | 1.74 | 3.48 | 0.348 | - | 99.7% |
| Fireworks (via OpenRouter) | 1.74 | 3.48 | 0.145 | - | 0% |
| Parasail (via OpenRouter)fp8 | 1.74 | 3.48 | 0.100 | - | 95.3% |
| Together AI (inference host) | 1.74 | 3.48 | 0.200 | - | - |
| Official - DeepSeek | 1.32 | 3.96 | 0.044 | - | - |
| Off-peakpeak 01:00 - 04:00 and 06:00 - 10:00 UTC, Mon-Fri; all other hours off-peak | 0.660 | 1.98 | 0.022 | - | - |
| TokenRouter (aggregator) | 1.32 | 3.96 | 0.044 | - | - |
| Azure (via OpenRouter) | 1.91 | 3.83 | 0.160 | - | 99.4% |
5.5× between the cheapest and the dearest of 22 sellers.
Against its tier
- This model$1.15
- Cheapest · GLM 5.3 Flash$0.29
- Tier median · Grok 4.6$8.00
- Dearest · GPT 5.5 Pro (04-23 snapshot)$330.00
#2 of 31 flagship · cheaper than 97%
Details
- Context window
- 1.0M1,048,576 tokens
- Developer
- DeepSeek
- Intelligence
- 45.3Reasoning, Max Effort
- Sellers
- 225.5× spread
- Cached input
- $0.003where the source publishes one
- Cheapest at
- Nahcrof (inference host)
- Combined
- $1.15in + out per 1M
- Last checked
- 2026-08-28
- Throughput
- 11 tok/sas published by Nahcrof (inference host)
- Availability
- 99.44%median of 17 OpenRouter endpoints, last 30 min