DeepSeek-V4-Pro
DeepSeek-V4-Pro is an open-weight language model from DeepSeek with 1.6T parameters (mixture of experts). At 4-bit it needs about 1056.4 GB of memory, more than any single device in our directory has.
Overview
Includes room for an 8k context and runtime overhead. More context needs more memory.
Key information
- Lab
- DeepSeek
- Released
- 22 Apr 2026
- Parameters
- 1.6T · MoE
- Type
- Language
- Context
- 1,048,576 tokens
- License
- MIT
- HF downloads, 30 days
- 319,188
- Repository
- deepseek-ai/DeepSeek-V4-Pro
More from DeepSeek
View allWill it run on your hardware?
How we estimate| Estimate or device | Memory |
|---|---|
| Q4 | 1056.4 GB |
| Q8 | 1865.4 GB |
| BF16 | 3518.6 GB |
| RTX 5060 Ti 16 GB | 16 GB usable |
| RTX 5090 32 GB | 32 GB usable |
| DGX Spark 128 GB | 108.8 GB usable |
| Mac Studio M5 Ultra 256 GB | 192 GB usable |
Single-device memory estimate. Pick a device to update the On your device column.
36 GB usable memory
| Format | Memory needed | On your device | Example hardware |
|---|---|---|---|
| 4-bit (Q4) | ≈ 1056.4 GB | Too big 2934% | No single device in our directory has enough memory |
| 8-bit (Q8) | ≈ 1865.4 GB | Too big 5182% | No single device in our directory has enough memory |
| 16-bit (BF16) | ≈ 3518.6 GB | Too big 9774% | No single device in our directory has enough memory |
Fits means the estimate stays within 90% of usable memory. Tight is above 90% and up to 100%. Too big is above 100%. These are memory checks, not speed or compatibility tests.
Mac allowances are 66% of total RAM up to 36 GB, and 75% above. AI boxes allow 85% because their memory is shared with the operating system.
Where to run it
On a Mac
Start with the MLX files, built for Apple Silicon.
mlx-community/DeepSeek-V4-ProCheck your MacThrough an API
StreamLake (fp8) is the cheapest provider we found.
$0.3631 / $0.7263
Input / output per 1M tokens
₹34 / ₹68 at ₹94.29 per $1
Blended: $0.4539 (₹43) per 1M. Ranked by a 3:1 mix of input and output tokens.
mlx_lm.generate --model mlx-community/DeepSeek-V4-Pro \
--prompt "Explain how a transformer works."Example commands built from the repository ids above. They need a working install of MLX. We have not run them; check each repository for available files.
| Precision | |||||
|---|---|---|---|---|---|
| StreamLakeCheapest | $0.4539₹43 | $0.3631₹34 | $0.7263₹68 | 1,024,000 | FP8 |
| Baidu | $0.4648₹44 | $0.3718₹35 | $0.7436₹70 | 1,048,576 | FP8 |
| Relace | $1.05₹99 | $0.90₹85 | $1.50₹141 | 1,048,576 | FP4 |
| SiliconFlow | $1.125₹106 | $1.00₹94 | $1.50₹141 | 1,048,576 | FP8 |
| GMICloud | $1.1963₹113 | $0.957₹90 | $1.914₹180 | 1,048,576 | FP8 |
| DigitalOcean | $1.305₹123 | $1.044₹98 | $2.088₹197 | 1,048,576 | Not listed |
| DeepInfra | $1.625₹153 | $1.30₹123 | $2.60₹245 | 1,048,576 | FP8 |
| Alibaba | $1.77₹167 | $1.416₹134 | $2.832₹267 | 1,000,000 | FP8 |
| Novita | $2.00₹189 | $1.60₹151 | $3.20₹302 | 1,048,576 | FP8 |
| Venice | $2.0627₹194 | $1.65₹156 | $3.301₹311 | 1,000,000 | Not listed |
| AtlasCloud | $2.105₹198 | $1.68₹158 | $3.38₹319 | 1,048,576 | FP4 |
| Azure | $2.39₹225 | $1.91₹180 | $3.83₹361 | 1,048,576 | Not listed |
INR at ₹94.29 per $1. Prices fetched 14 Nov 2026.
More about this model
Frequently asked questions
How much memory does DeepSeek-V4-Pro need? At an 8k context, about 1056.4 GB at 4-bit, 1865.4 GB at 8-bit and 3518.6 GB at 16-bit (BF16). A longer context needs more.
Can I run DeepSeek-V4-Pro on a Mac? No Mac in our directory has enough usable memory for the 4-bit estimate of 1056.4 GB. An MLX build is available: mlx-community/DeepSeek-V4-Pro.
Does the full context fit? DeepSeek-V4-Pro supports up to 1,048,576 tokens of context. Our memory estimates use 8k, so long prompts need more memory than shown.
What does it cost through an API? The cheapest provider we found is StreamLake, at $0.36 per 1M input tokens and $0.73 per 1M output tokens (₹34 and ₹68 at 94.29 rupees per dollar). Ranked by a 3:1 mix of input and output tokens.
License in plain English
MIT. This license permits commercial use, modification and redistribution, as long as the copyright and license notice are kept. This summary is not legal advice.
OMHQ RigBench
Not yet tested. We show no speed or quality score without a measurement.
Revision history and sources
Last updated . Sources:Hugging Face model card, MLX repository, OpenRouter.