DeepSeek V4 Flash vs Qwen3.5 397B A17B

DeepSeek V4 Flash leads llmrun Score 115.3 to 112.8 (ranges overlap); neither fits 24 GB.

Run DeepSeek V4 Flash in the cloud →

Model A · DeepSeek

DeepSeek V4 Flash

115.3llmrun Score

290.9B · 14.1B active · 1M context · MIT · Apr 2026

Chat

Model B · Alibaba

Qwen3.5 397B A17B

112.8llmrun Score

403.4B · 24.4B active · 256K context · Apache 2.0 · Feb 2026

VisionFunctions

Which should you choose?

Choose DeepSeek V4 Flash if you want…

Higher llmrun Score
115.3 vs 112.8, ranges overlap
Lighter on memory
174.9 vs 242.5 GB at Q4_K_M
Longer context window
1M vs 256K tokens
Also does chat
Qwen3.5 397B A17B doesn't list it

Choose Qwen3.5 397B A17B if you want…

Also does vision, function-calling
DeepSeek V4 Flash doesn't list it

On your hardware

Each model at its default quant (Q4_K_M), a smaller one when it must. Estimates, ±20%.

MemoryDeepSeek V4 FlashQwen3.5 397B A17B

8 GB

RTX 4060 · 8 GB tier

DeepSeek V4 Flash

Doesn't fit

smallest quant needs 80.5 GB

Qwen3.5 397B A17B

Doesn't fit

smallest quant needs 111.5 GB

12 GB

RTX 3060 12GB · 12 GB tier

DeepSeek V4 Flash

Doesn't fit

smallest quant needs 80.5 GB

Qwen3.5 397B A17B

Doesn't fit

smallest quant needs 111.5 GB

16 GB

RTX 5060 Ti 16GB · 16 GB tier

DeepSeek V4 Flash

Doesn't fit

smallest quant needs 80.5 GB

Qwen3.5 397B A17B

Doesn't fit

smallest quant needs 111.5 GB

24 GBreference

RTX 4090 · 24 GB tier

DeepSeek V4 Flash

Doesn't fit

smallest quant needs 80.5 GB

Qwen3.5 397B A17B

Doesn't fit

smallest quant needs 111.5 GB

32 GB

RTX 5090 · 32 GB tier

DeepSeek V4 Flash

Doesn't fit

smallest quant needs 80.5 GB

Qwen3.5 397B A17B

Doesn't fit

smallest quant needs 111.5 GB

48 GB

RTX A6000 · 48 GB tier

DeepSeek V4 Flash

Doesn't fit

smallest quant needs 80.5 GB

Qwen3.5 397B A17B

Doesn't fit

smallest quant needs 111.5 GB

Mac 32 GB

Mac Mini M4 · Macs

DeepSeek V4 Flash

Doesn't fit

smallest quant needs 80.5 GB

Qwen3.5 397B A17B

Doesn't fit

smallest quant needs 111.5 GB

Mac 64 GB

Mac Studio M4 Max · Macs

DeepSeek V4 Flash

Doesn't fit

smallest quant needs 80.5 GB

Qwen3.5 397B A17B

Doesn't fit

smallest quant needs 111.5 GB

Mac 128 GB

Mac Studio M4 Max · Macs

DeepSeek V4 Flash

Q2_K · ~41 tok/s

124.1 GB · up to 32K

Qwen3.5 397B A17B

IQ2_XS · ~32 tok/s

121.6 GB · up to 128K

Reference GPU: RTX 4090. Tinted cell = the better pick on that machine (fits, more headroom, or at least 10% faster).

llmrun Score

with its 90% range · bars share one scale

DeepSeek V4 Flash Qwen3.5 397B A17B

Overall

DeepSeek V4 Flash

115.3111–119

Qwen3.5 397B A17B

112.8109–117

Coding

DeepSeek V4 Flash

114.3109–120

Qwen3.5 397B A17B

no data

Math

DeepSeek V4 Flash

no data

Qwen3.5 397B A17B

112.4107–118

Science

DeepSeek V4 Flash

111.292–131

Qwen3.5 397B A17B

113.1102–124

Reasoning

DeepSeek V4 Flash

110.9105–117

Qwen3.5 397B A17B

116.7108–125

DeepSeek V4 Flash · #24 of 106 open models · #8 of 82 that fit ≤192 GB

Qwen3.5 397B A17B · #26 of 106 open models · #13 of 89 that fit ≤256 GB

Benchmarks

3 shared boards: DeepSeek V4 Flash leads 1, Qwen3.5 397B A17B leads 2 · Benchmarks imported Nov 2026

BoardDeepSeek V4 FlashQwen3.5 397B A17B

CritPt

DeepSeek V4 Flash

7.1%#19/84 open

Lab-reportedEpoch AI · Apr 2026

Qwen3.5 397B A17B

0.9%#42/84 open

Lab-reportedEpoch AI · Feb 2026

DTBench

DeepSeek V4 Flash

86.4%#15/70 open

Lab-reportedEpoch AI · Apr 2026

Qwen3.5 397B A17B

87.5%#13/70 open

Lab-reportedEpoch AI · Feb 2026

LMCA

DeepSeek V4 Flash

35.9%#14/56 open

Lab-reportedEpoch AI · Apr 2026

Qwen3.5 397B A17B

37.9%#10/56 open

Lab-reportedEpoch AI · Feb 2026

22 more boards (other shared and single-model boards)

LMArena Text

DeepSeek V4 Flash

1438.7#21/188 open

ArenaLMArena · Oct 2026

Qwen3.5 397B A17B

1441.4#18/188 open

ArenaLMArena · Oct 2026

LMArena WebDev

DeepSeek V4 Flash

1581.8#14/55 open

ArenaLMArena · Oct 2026

Qwen3.5 397B A17B

1400.2#30/55 open

ArenaLMArena · Oct 2026

Only DeepSeek V4 Flash (14)

Only Qwen3.5 397B A17B (6)

VRAM by quantization

Memory to load the model and 2K context; below it, the total at 32K and 128K context. Lighter side in bold.

QuantDeepSeek V4 FlashQwen3.5 397B A17B

Q2_K

DeepSeek V4 Flash

124 GB107.0 GB file

32K 124 GB · 128K 125 GB

Qwen3.5 397B A17B

172 GB146.3 GB file

32K 172 GB · 128K 174 GB

Q3_K_S

DeepSeek V4 Flash

not listed

Qwen3.5 397B A17B

177 GB164.3 GB file

32K 177 GB · 128K 179 GB

Q3_K_M

DeepSeek V4 Flash

142 GB129.3 GB file

32K 143 GB · 128K 144 GB

Qwen3.5 397B A17B

197 GB177.4 GB file

32K 197 GB · 128K 199 GB

Q4_0

DeepSeek V4 Flash

not listed

Qwen3.5 397B A17B

202 GB~201.7 GB file

32K 203 GB · 128K 204 GB

Q4_K_M

DeepSeek V4 Flash

175 GB174.9 GB file

32K 175 GB · 128K 176 GB

Qwen3.5 397B A17B

242 GB244.1 GB file

32K 243 GB · 128K 244 GB

Q5_K_M

DeepSeek V4 Flash

208 GB~207.3 GB file

32K 208 GB · 128K 209 GB

Qwen3.5 397B A17B

288 GB293.7 GB file

32K 288 GB · 128K 290 GB

Q6_K

DeepSeek V4 Flash

240 GB~240.0 GB file

32K 241 GB · 128K 242 GB

Qwen3.5 397B A17B

333 GB326.6 GB file

32K 334 GB · 128K 335 GB

Q8_0

DeepSeek V4 Flash

291 GB302.3 GB file

32K 292 GB · 128K 293 GB

Qwen3.5 397B A17B

404 GB421.5 GB file

32K 404 GB · 128K 406 GB

18 more quants

IQ2_XXS

DeepSeek V4 Flash

80.3 GB90.9 GB file

32K 80.7 GB · 128K 81.8 GB

Qwen3.5 397B A17B

111 GB114.9 GB file

32K 112 GB · 128K 113 GB

IQ2_XS

DeepSeek V4 Flash

87.6 GB87.0 GB file

32K 88.0 GB · 128K 89.0 GB

Qwen3.5 397B A17B

121 GB125.4 GB file

32K 122 GB · 128K 123 GB

IQ2_S

DeepSeek V4 Flash

91.2 GB88.0 GB file

32K 91.6 GB · 128K 92.7 GB

Qwen3.5 397B A17B

126 GB127.3 GB file

32K 127 GB · 128K 128 GB

IQ2_M

DeepSeek V4 Flash

98.5 GB90.9 GB file

32K 98.9 GB · 128K 99.9 GB

Qwen3.5 397B A17B

137 GB123.1 GB file

32K 137 GB · 128K 138 GB

IQ3_XXS

DeepSeek V4 Flash

113 GB103.0 GB file

32K 113 GB · 128K 114 GB

Qwen3.5 397B A17B

157 GB140.3 GB file

32K 157 GB · 128K 159 GB

Q2_K_S

DeepSeek V4 Flash

117 GB98.6 GB file

32K 117 GB · 128K 118 GB

Qwen3.5 397B A17B

not listed

IQ3_XS

DeepSeek V4 Flash

not listed

Qwen3.5 397B A17B

167 GB188.2 GB file

32K 167 GB · 128K 169 GB

IQ3_S

DeepSeek V4 Flash

124 GB117.3 GB file

32K 124 GB · 128K 125 GB

Qwen3.5 397B A17B

172 GB146.4 GB file

32K 172 GB · 128K 174 GB

IQ3_M

DeepSeek V4 Flash

131 GB136.1 GB file

32K 132 GB · 128K 133 GB

Qwen3.5 397B A17B

182 GB196.0 GB file

32K 182 GB · 128K 184 GB

Q3_K_L

DeepSeek V4 Flash

not listed

Qwen3.5 397B A17B

207 GB195.9 GB file

32K 208 GB · 128K 209 GB

IQ4_XS

DeepSeek V4 Flash

157 GB137.9 GB file

32K 157 GB · 128K 158 GB

Qwen3.5 397B A17B

217 GB189.7 GB file

32K 218 GB · 128K 219 GB

IQ4_NL

DeepSeek V4 Flash

164 GB137.9 GB file

32K 164 GB · 128K 165 GB

Qwen3.5 397B A17B

227 GB193.8 GB file

32K 228 GB · 128K 229 GB

Q4_1

DeepSeek V4 Flash

not listed

Qwen3.5 397B A17B

227 GB256.1 GB file

32K 228 GB · 128K 229 GB

Q4_K_S

DeepSeek V4 Flash

not listed

Qwen3.5 397B A17B

227 GB228.0 GB file

32K 228 GB · 128K 229 GB

Q4_K_L

DeepSeek V4 Flash

not listed

Qwen3.5 397B A17B

248 GB249.0 GB file

32K 248 GB · 128K 249 GB

Q5_K_S

DeepSeek V4 Flash

not listed

Qwen3.5 397B A17B

278 GB276.6 GB file

32K 278 GB · 128K 280 GB

Q5_K_L

DeepSeek V4 Flash

not listed

Qwen3.5 397B A17B

293 GB290.6 GB file

32K 293 GB · 128K 295 GB

BF16

DeepSeek V4 Flash

582 GB~581.9 GB file

32K 583 GB · 128K 584 GB

Qwen3.5 397B A17B

807 GB793.0 GB file

32K 808 GB · 128K 809 GB

How to run each

DeepSeek V4 Flash

Ollama · Q4_K_M

$ollama run hf.co/teamblobfish/DeepSeek-V4-Flash-GGUF:Q4_K_M

All ways to run DeepSeek V4 Flash →

Qwen3.5 397B A17B

Ollama · Q4_K_M

$ollama run hf.co/unsloth/Qwen3.5-397B-A17B-GGUF:Q4_K_M

All ways to run Qwen3.5 397B A17B →

Specifications

Specifications of DeepSeek V4 Flash and Qwen3.5 397B A17B
Spec	DeepSeek V4 Flash	Qwen3.5 397B A17B
Publisher	DeepSeek	Alibaba
Parameters	290.9B	403.4B
Active parameters	14B per token	24B per token
Architecture	DeepseekV4ForCausalLM	Qwen3_5MoeForConditionalGeneration
Context	1M tokens	256K tokens
Modalities	Chat	Vision, Function-calling
License	MIT	Apache 2.0
Released	Apr 2026	Feb 2026

Frequently Asked Questions

Which is better, DeepSeek V4 Flash or Qwen3.5 397B A17B?

DeepSeek V4 Flash leads llmrun Score 115.3 to 112.8 (ranges overlap); neither fits 24 GB. Choose DeepSeek V4 Flash for higher llmrun Score and lighter on memory; choose Qwen3.5 397B A17B for also does vision, function-calling.

Which needs less VRAM, DeepSeek V4 Flash or Qwen3.5 397B A17B?

At Q4_K_M, DeepSeek V4 Flash requires 174.9 GB while Qwen3.5 397B A17B needs 242.5 GB, making DeepSeek V4 Flash the easier model to run on your local setup.

Can I run DeepSeek V4 Flash or Qwen3.5 397B A17B on a 16 GB GPU?

DeepSeek V4 Flash does not fit (smallest quant needs 80.5 GB); Qwen3.5 397B A17B does not fit (smallest quant needs 111.5 GB).

Which has a longer context window, DeepSeek V4 Flash or Qwen3.5 397B A17B?

DeepSeek V4 Flash supports 1,048,576 tokens and Qwen3.5 397B A17B supports 262,144 tokens.