llmrun

DeepSeek V4 Pro vs Qwen3.8 27B

Qwen3.8 27B leads llmrun Score 122.1 to 116.5; only Qwen3.8 27B fits 24 GB (Q4_K_M); DeepSeek V4 Pro needs 680 GB even at its smallest quant.

Model A · DeepSeek

DeepSeek V4 Pro
116.5llmrun Score

1598.8B · 50.7B active · 1M context · MIT · Apr 2026

Chat

Model B · Alibaba

Qwen3.8 27B
122.1llmrun Score

27.8B · 256K context · Apache 2.0 · Aug 2026

VisionFunctions

Which should you choose?

Choose DeepSeek V4 Pro if you want…

  • Stronger at science

    122.3 vs 112.0

  • Longer context window

    1M vs 256K tokens

  • Also does chat

    Qwen3.8 27B doesn't list it

Choose Qwen3.8 27B if you want…

  • Higher llmrun Score

    122.1 vs 116.5

  • Stronger at coding

    119.7 vs 110.6

  • Fits a 12 GB GPU

    IQ3_XXS · 12 GB; DeepSeek V4 Pro needs 680 GB at its smallest quant

  • Fits a Mac 32 GB

    Q4_K_M · 17.9 GB; DeepSeek V4 Pro needs 680 GB at its smallest quant

  • Lighter on memory

    17.4 vs 959.6 GB at Q4_K_M

  • Also does vision, function-calling

    DeepSeek V4 Pro doesn't list it

On your hardware

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

8 GB

RTX 4060 · 8 GB tier

DeepSeek V4 Pro

Doesn't fit

smallest quant needs 680 GB

Qwen3.8 27B

Doesn't fit

smallest quant needs 8.8 GB

12 GB

RTX 3060 12GB · 12 GB tier

DeepSeek V4 Pro

Doesn't fit

smallest quant needs 680 GB

Qwen3.8 27B

IQ3_XXS · ~20 tok/s

12 GB · up to 16Kfits

16 GB

RTX 5060 Ti 16GB · 16 GB tier

DeepSeek V4 Pro

Doesn't fit

smallest quant needs 680 GB

Qwen3.8 27B

Q3_K_L · ~19 tok/s

15.4 GB · up to 16Kfits

24 GBreference

RTX 4090 · 24 GB tier

DeepSeek V4 Pro

Doesn't fit

smallest quant needs 680 GB

Qwen3.8 27B

Q4_K_M · ~38 tok/s

17.9 GB · up to 64Kfits

32 GB

RTX 5090 · 32 GB tier

DeepSeek V4 Pro

Doesn't fit

smallest quant needs 680 GB

Qwen3.8 27B

Q4_K_M · ~67 tok/s

17.9 GB · up to 256Kfits

48 GB

RTX A6000 · 48 GB tier

DeepSeek V4 Pro

Doesn't fit

smallest quant needs 680 GB

Qwen3.8 27B

Q4_K_M · ~29 tok/s

17.9 GB · up to 256Kfits

Mac 32 GB

Mac Mini M4 · Macs

DeepSeek V4 Pro

Doesn't fit

smallest quant needs 680 GB

Qwen3.8 27B

Q4_K_M · ~5 tok/s

17.9 GB · up to 128Kfits

Mac 64 GB

Mac Studio M4 Max · Macs

DeepSeek V4 Pro

Doesn't fit

smallest quant needs 680 GB

Qwen3.8 27B

Q4_K_M · ~22 tok/s

17.9 GB · up to 256Kfits

Mac 128 GB

Mac Studio M4 Max · Macs

DeepSeek V4 Pro

Doesn't fit

smallest quant needs 680 GB

Qwen3.8 27B

Q4_K_M · ~22 tok/s

17.9 GB · up to 256Kfits

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

Overall

DeepSeek V4 Pro

116.5114–119

Qwen3.8 27B

122.1120–124

Coding

DeepSeek V4 Pro

110.6105–116

Qwen3.8 27B

119.7114–125

Agents

DeepSeek V4 Pro

no data

Qwen3.8 27B

122.8117–129

Math

DeepSeek V4 Pro

117.6114–122

Qwen3.8 27B

117.5111–124

Science

DeepSeek V4 Pro

122.3111–134

Qwen3.8 27B

112.093–131

Reasoning

DeepSeek V4 Pro

117.0110–124

Qwen3.8 27B

124.0121–127

DeepSeek V4 Pro · #16 of 100 open models · #16 of 99 that fit ≤1 TB

Qwen3.8 27B · #11 of 100 open models · #1 of 35 that fit 24 GB

Benchmarks

9 shared boards: DeepSeek V4 Pro leads 5, Qwen3.8 27B leads 2 · Benchmarks imported Oct 2026

LiveBench Coding

DeepSeek V4 Pro

70.0#16/19 open

IndependentLiveBench · Jun 2026

Qwen3.8 27B

75.7#9/19 open

IndependentLiveBench · Jun 2026

LiveBench Reasoning

DeepSeek V4 Pro

82.7#10/19 open

IndependentLiveBench · Jun 2026

Qwen3.8 27B

80.0#11/19 open

IndependentLiveBench · Jun 2026

LiveBench Math

DeepSeek V4 Pro

90.7#3/19 open

IndependentLiveBench · Jun 2026

Qwen3.8 27B

86.2#11/19 open

IndependentLiveBench · Jun 2026

SciCode

DeepSeek V4 Pro

50.0%#12/66 open

Lab-reportedEpoch AI · Apr 2026

Qwen3.8 27B

46.6%#19/66 open

Lab-reportedEpoch AI · Aug 2026

CritPt

DeepSeek V4 Pro

12.9%#9/66 open

Lab-reportedEpoch AI · Apr 2026

Qwen3.8 27B

5.4%#15/66 open

Lab-reportedEpoch AI · Aug 2026

DTBench

DeepSeek V4 Pro

90.7%#6/68 open

Lab-reportedEpoch AI · Apr 2026

Qwen3.8 27B

88.0%#9/68 open

Lab-reportedEpoch AI · Aug 2026

LMCA

DeepSeek V4 Pro

41.2%#8/54 open

Lab-reportedEpoch AI · Apr 2026

Qwen3.8 27B

41.4%#7/54 open

Lab-reportedEpoch AI · Aug 2026

ProofBench

DeepSeek V4 Pro

16.0%#14/26 open

Lab-reportedEpoch AI · Apr 2026

Qwen3.8 27B

16.0%#16/26 open

Lab-reportedEpoch AI · Aug 2026

Surface Evolver Bench

DeepSeek V4 Pro

40.0%#9/13 open

Lab-reportedEpoch AI · Apr 2026

Qwen3.8 27B

45.0%#7/13 open

Lab-reportedEpoch AI · Aug 2026

25 more boards (other shared and single-model boards)
LMArena Text

DeepSeek V4 Pro

1457.8#11/187 open

ArenaLMArena · Oct 2026

Qwen3.8 27B

1437.8#22/187 open

ArenaLMArena · Oct 2026

LMArena WebDev

DeepSeek V4 Pro

1464.2#20/52 open

ArenaLMArena · Oct 2026

Qwen3.8 27B

1593.2#10/52 open

ArenaLMArena · Oct 2026

VRAM by quantization

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

Q2_K

DeepSeek V4 Pro

680 GB574.1 GB file

32K 680 GB · 128K 682 GB

Qwen3.8 27B

12.6 GB10.8 GB file

32K 13.9 GB · 128K 19.3 GB

Q3_K_S

DeepSeek V4 Pro

not listed

Qwen3.8 27B

12.9 GB12.7 GB file

32K 14.2 GB · 128K 19.6 GB

Q3_K_M

DeepSeek V4 Pro

780 GB748.7 GB file

32K 780 GB · 128K 782 GB

Qwen3.8 27B

14.3 GB13.4 GB file

32K 15.6 GB · 128K 21.0 GB

Q4_0

DeepSeek V4 Pro

not listed

Qwen3.8 27B

14.6 GB~13.9 GB file

32K 16.0 GB · 128K 21.4 GB

Q4_K_M

DeepSeek V4 Pro

960 GB954.6 GB file

32K 960 GB · 128K 961 GB

Qwen3.8 27B

17.4 GB16.5 GB file

32K 18.8 GB · 128K 24.1 GB

Q5_K_M

DeepSeek V4 Pro

1140 GB~1139.2 GB file

32K 1140 GB · 128K 1141 GB

Qwen3.8 27B

20.5 GB19.8 GB file

32K 21.9 GB · 128K 27.3 GB

Q6_K

DeepSeek V4 Pro

1319 GB~1319.0 GB file

32K 1320 GB · 128K 1321 GB

Qwen3.8 27B

23.7 GB22.0 GB file

32K 25.0 GB · 128K 30.4 GB

Q8_0

DeepSeek V4 Pro

1599 GB1671.8 GB file

32K 1600 GB · 128K 1601 GB

Qwen3.8 27B

28.5 GB29.0 GB file

32K 29.9 GB · 128K 35.2 GB

17 more quants

IQ2_XXS

DeepSeek V4 Pro

not listed

Qwen3.8 27B

8.4 GB7.3 GB file

32K 9.7 GB · 128K 15.1 GB

IQ2_XS

DeepSeek V4 Pro

not listed

Qwen3.8 27B

9.1 GB9.1 GB file

32K 10.4 GB · 128K 15.8 GB

IQ2_S

DeepSeek V4 Pro

not listed

Qwen3.8 27B

9.4 GB8.4 GB file

32K 10.8 GB · 128K 16.1 GB

IQ2_M

DeepSeek V4 Pro

not listed

Qwen3.8 27B

10.1 GB10.5 GB file

32K 11.5 GB · 128K 16.8 GB

IQ3_XXS

DeepSeek V4 Pro

not listed

Qwen3.8 27B

11.5 GB10.9 GB file

32K 12.9 GB · 128K 18.2 GB

IQ3_XS

DeepSeek V4 Pro

not listed

Qwen3.8 27B

12.2 GB10.4 GB file

32K 13.6 GB · 128K 18.9 GB

IQ3_S

DeepSeek V4 Pro

not listed

Qwen3.8 27B

12.6 GB12.0 GB file

32K 13.9 GB · 128K 19.3 GB

IQ3_M

DeepSeek V4 Pro

not listed

Qwen3.8 27B

13.3 GB14.9 GB file

32K 14.6 GB · 128K 20.0 GB

Q3_K_L

DeepSeek V4 Pro

not listed

Qwen3.8 27B

15.0 GB14.1 GB file

32K 16.3 GB · 128K 21.7 GB

IQ4_XS

DeepSeek V4 Pro

not listed

Qwen3.8 27B

15.7 GB14.3 GB file

32K 17.0 GB · 128K 22.4 GB

IQ4_NL

DeepSeek V4 Pro

not listed

Qwen3.8 27B

16.4 GB17.4 GB file

32K 17.7 GB · 128K 23.1 GB

Q4_1

DeepSeek V4 Pro

not listed

Qwen3.8 27B

16.4 GB17.5 GB file

32K 17.7 GB · 128K 23.1 GB

Q4_K_S

DeepSeek V4 Pro

not listed

Qwen3.8 27B

16.4 GB15.4 GB file

32K 17.7 GB · 128K 23.1 GB

Q4_K_L

DeepSeek V4 Pro

not listed

Qwen3.8 27B

17.8 GB18.8 GB file

32K 19.1 GB · 128K 24.5 GB

Q5_K_S

DeepSeek V4 Pro

not listed

Qwen3.8 27B

19.9 GB18.7 GB file

32K 21.2 GB · 128K 26.6 GB

Q5_K_L

DeepSeek V4 Pro

not listed

Qwen3.8 27B

20.9 GB21.5 GB file

32K 22.2 GB · 128K 27.6 GB

BF16

DeepSeek V4 Pro

3198 GB~3197.7 GB file

32K 3198 GB · 128K 3200 GB

Qwen3.8 27B

56.3 GB54.7 GB file

32K 57.7 GB · 128K 63.0 GB

How to run each

Specifications

Specifications of DeepSeek V4 Pro and Qwen3.8 27B
SpecDeepSeek V4 ProQwen3.8 27B
PublisherDeepSeekAlibaba
Parameters1598.8B27.8B
Active parameters51B per token—
ArchitectureDeepseekV4ForCausalLMQwen3_5ForConditionalGeneration
Context1M tokens256K tokens
ModalitiesChatVision, Function-calling
LicenseMITApache 2.0
ReleasedApr 2026Aug 2026

Frequently Asked Questions

Which is better, DeepSeek V4 Pro or Qwen3.8 27B?

Qwen3.8 27B leads llmrun Score 122.1 to 116.5; only Qwen3.8 27B fits 24 GB (Q4_K_M); DeepSeek V4 Pro needs 680 GB even at its smallest quant. Choose DeepSeek V4 Pro for stronger at science and longer context window; choose Qwen3.8 27B for higher llmrun Score and stronger at coding.

Which needs less VRAM, DeepSeek V4 Pro or Qwen3.8 27B?

At Q4_K_M, DeepSeek V4 Pro requires 959.6 GB while Qwen3.8 27B needs 17.4 GB, making Qwen3.8 27B the easier model to run on your local setup.

Can I run DeepSeek V4 Pro or Qwen3.8 27B on a 16 GB GPU?

DeepSeek V4 Pro does not fit (smallest quant needs 680 GB); Qwen3.8 27B fits at Q3_K_L.

Which has a longer context window, DeepSeek V4 Pro or Qwen3.8 27B?

DeepSeek V4 Pro supports 1,048,576 tokens and Qwen3.8 27B supports 262,144 tokens.

DeepSeek V4 Pro vs Qwen3.8 27B — VRAM & GPU Comparison | llmrun