DeepSeek-V4-Flash
DeepSeek-V4-Flash is an open-weight language model from DeepSeek with 291B parameters (mixture of experts). At 4-bit it needs about 192.8 GB of memory, so the smallest device in our directory that fits it is the Rubin, which is announced but not shipping yet.
Overview
Includes room for an 8k context and runtime overhead. More context needs more memory.
Key information
- Lab
- DeepSeek
- Released
- 22 Apr 2026
- Parameters
- 291B · MoE
- Type
- Language
- Context
- 1,048,576 tokens
- License
- MIT
- HF downloads, 30 days
- 1,206,494
- Repository
- deepseek-ai/DeepSeek-V4-Flash
More from DeepSeek
View allWill it run on your hardware?
How we estimate| Estimate or device | Memory |
|---|---|
| Q4 | 192.8 GB |
| Q8 | 340 GB |
| BF16 | 640.9 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) | ≈ 192.8 GB | Too big 536% | RubinAnnounced288 GB memory |
| 8-bit (Q8) | ≈ 340 GB | Too big 944% | Mac Studio M3 Ultra (512GB)Discontinued512 GB memory |
| 16-bit (BF16) | ≈ 640.9 GB | Too big 1780% | 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
With GGUF
Choose a file for llama.cpp or a compatible app.
unsloth/DeepSeek-V4-Flash-GGUFFind compatible hardwareThrough an API
Baidu (fp8) is the cheapest provider we found.
$0.0251 / $0.0501
Input / output per 1M tokens
₹2.37 / ₹4.72 at ₹94.29 per $1
Blended: $0.0314 (₹2.96) per 1M. Ranked by a 3:1 mix of input and output tokens.
llama-cli -hf unsloth/DeepSeek-V4-Flash-GGUF \
-c 8192 -p "Explain how a transformer works."Example commands built from the repository ids above. They need a working install of llama.cpp. We have not run them; check each repository for available files.
| Precision | |||||
|---|---|---|---|---|---|
| BaiduCheapest | $0.0314₹2.96 | $0.0251₹2.37 | $0.0501₹4.72 | 1,048,576 | FP8 |
| StreamLake | $0.035₹3.30 | $0.028₹2.64 | $0.056₹5.28 | 1,024,000 | FP8 |
| DeepInfra | $0.1125₹11 | $0.09₹8.49 | $0.18₹17 | 1,048,576 | FP8 |
| GMICloud | $0.1138₹11 | $0.091₹8.58 | $0.182₹17 | 1,048,575 | FP8 |
| Venice | $0.1206₹11 | $0.0966₹9.11 | $0.1925₹18 | 1,000,000 | Not listed |
| Fireworks | $0.1313₹12 | $0.105₹9.90 | $0.21₹20 | 1,048,576 | Not listed |
| Relace | $0.1313₹12 | $0.0251₹2.37 | $0.45₹42 | 1,048,576 | FP4 |
| Wafer | $0.1325₹12 | $0.12₹11 | $0.17₹16 | 1,048,576 | MXFP4 |
| SiliconFlow | $0.1675₹16 | $0.13₹12 | $0.28₹26 | 1,048,576 | FP8 |
| AtlasCloud | $0.175₹17 | $0.14₹13 | $0.28₹26 | 1,048,576 | FP4 |
| Parasail | $0.175₹17 | $0.14₹13 | $0.28₹26 | 1,048,576 | FP8 |
| Mancer 2 | $0.2375₹22 | $0.15₹14 | $0.50₹47 | 1,048,576 | FP8 |
| Azure | $0.2975₹28 | $0.21₹20 | $0.56₹53 | 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-Flash need? At an 8k context, about 192.8 GB at 4-bit, 340 GB at 8-bit and 640.9 GB at 16-bit (BF16). A longer context needs more.
Can I run DeepSeek-V4-Flash on a Mac? No current Mac fits the 4-bit estimate of 192.8 GB. The smallest Mac in our directory that does is the Mac Studio M3 Ultra (512GB), which is discontinued. We have not found a trusted MLX build yet, but a GGUF build is available.
Does the full context fit? DeepSeek-V4-Flash 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 Baidu, at $0.03 per 1M input tokens and $0.05 per 1M output tokens (₹2.37 and ₹4.72 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, GGUF repository, OpenRouter.