DeepSeek

DeepSeek V4 Flash 0731

DeepSeek V4 Flash 0731 is an open-weights 162B-parameter model by DeepSeek. It holds an Intelligence score of 77.5 on the local.ai leaderboard (2026-08 snapshot).

open weights
77.5
local.ai Intelligence
Parameters
162B
Benchmarks (local.ai snapshot)
τ²-bench
85%
GAIA
76%
GDPval
77%
HuggingFace metadata · updated 2026-08-01
repo
deepseek-ai/DeepSeek-V4-Flash-0731
license
mit
downloads
1,798,247
likes
3,411

Metadata refreshed from the HuggingFace API (snapshot 2026-08-15T15:53:42.405Z). Context length and long-form descriptions are the next pipeline stage.

Does it fit your hardware?

Detect or select your GPU first. Green = weights + context headroom, yellow = tight, grey = does not fit.

Q4_K_M 94GQ6_K 145.8GQ8_0 178.2GFP16 324G
Quantization≈ File sizeNote
IQ2_XXS43.7 GBExtreme compression, largest quality loss
IQ3_XXS58.3 GBVery small
Q3_K_M74.5 GBSmall
NVFP481 GBNVIDIA 4-bit
Q4_K_S87.5 GBCommon
Q4_K_M94 GBMost common sweet spot
Q5_K_M123.1 GBBalanced
Q6_K145.8 GBNear-lossless
Q8_0178.2 GBNear-lossless
FP16324 GBOriginal weights

Actual files on HuggingFace

FileSize
model-00048-of-00048.safetensors3.44 GB
model-00046-of-00048.safetensors3.36 GB
model-00004-of-00048.safetensors3.35 GB
model-00012-of-00048.safetensors3.34 GB
model-00014-of-00048.safetensors3.34 GB
model-00016-of-00048.safetensors3.34 GB
GGUF: unsloth/DeepSeek-V4-Flash-0731-GGUF · 306,443 downloads
FileSize
UD-IQ1_M/DeepSeek-V4-Flash-0731-UD-IQ1_M-00001-of-00003.gguf0.00 GBdownload
UD-IQ1_M/DeepSeek-V4-Flash-0731-UD-IQ1_M-00002-of-00003.gguf46.05 GBdownload
UD-IQ1_M/DeepSeek-V4-Flash-0731-UD-IQ1_M-00003-of-00003.gguf34.88 GBdownload
UD-IQ1_S/DeepSeek-V4-Flash-0731-UD-IQ1_S-00001-of-00003.gguf0.00 GBdownload
UD-IQ1_S/DeepSeek-V4-Flash-0731-UD-IQ1_S-00002-of-00003.gguf45.72 GBdownload
UD-IQ1_S/DeepSeek-V4-Flash-0731-UD-IQ1_S-00003-of-00003.gguf31.14 GBdownload
UD-IQ2_M/DeepSeek-V4-Flash-0731-UD-IQ2_M-00001-of-00003.gguf0.00 GBdownload
UD-IQ2_M/DeepSeek-V4-Flash-0731-UD-IQ2_M-00002-of-00003.gguf46.53 GBdownload
UD-IQ2_M/DeepSeek-V4-Flash-0731-UD-IQ2_M-00003-of-00003.gguf38.15 GBdownload
UD-IQ2_XXS/DeepSeek-V4-Flash-0731-UD-IQ2_XXS-00001-of-00003.gguf0.00 GBdownload
UD-IQ2_XXS/DeepSeek-V4-Flash-0731-UD-IQ2_XXS-00002-of-00003.gguf46.46 GBdownload
UD-IQ2_XXS/DeepSeek-V4-Flash-0731-UD-IQ2_XXS-00003-of-00003.gguf38.15 GBdownload
GGUF: huihui-ai/Huihui-DeepSeek-V4-Flash-0731-abliterated-GGUF · 243,479 downloads
FileSize
DeepSeek-V4-Flash-Q2-0731.gguf80.76 GBdownload
DeepSeek-V4-Flash-Q2_K-0731.gguf90.89 GBdownload
DeepSeek-V4-Flash-Q4-mxfp4-0731.gguf145.26 GBdownload
DeepSeek-V4-Flash-Q4_K-0731.gguf153.33 GBdownload
dspark-abliterated/dspark-DeepSeek-V4-Flash-0731-BF16.gguf10.54 GBdownload
dspark-abliterated/dspark-DeepSeek-V4-Flash-0731-Q8_0.gguf10.15 GBdownload
dspark/dspark-DeepSeek-V4-Flash-0731-BF16.gguf10.54 GBdownload
dspark/dspark-DeepSeek-V4-Flash-0731-Q8_0.gguf10.15 GBdownload

What you save self-hosting this model

Uses the model's own API price when it exists; otherwise the closest commercial equivalent.

Waiting for hardware detection…

RunLocal presents third-party leaderboard data (local.ai / Exo Labs, 2026-08) with its own fit and cost estimates. Verify pricing, licenses and file sizes at the source before deploying a model.