DeepSeek V4 Pro vs Mistral Large 4 1T A52B

DeepSeek V4 Pro and Mistral Large 4 1T A52B match in llmrun Score, with values of 116.7 versus 118.0; neither fits 24 GB.

Run Mistral Large 4 1T A52B in the cloud →

Model A · DeepSeek

DeepSeek V4 Pro

116.7llmrun Score

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

Chat

Model B · Mistral AI

Mistral Large 4 1T A52B

118.0llmrun Score

1000B · 52B active · 1M context · Other · Sep 2026

VisionFunctions

Which should you choose?

Choose DeepSeek V4 Pro if you want…

Also does chat
Mistral Large 4 1T A52B doesn't list it
More permissive license
MIT vs Other

Choose Mistral Large 4 1T A52B if you want…

Lighter on memory
603.9 vs 959.6 GB at Q4_K_M
Also does vision, function-calling
DeepSeek V4 Pro doesn't list it
Newer release
Sep 2026 vs Apr 2026

On your hardware

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

MemoryDeepSeek V4 ProMistral Large 4 1T A52B

8 GB

RTX 4060 · 8 GB tier

DeepSeek V4 Pro

Doesn't fit

smallest quant needs 680 GB

Mistral Large 4 1T A52B

Doesn't fit

smallest quant needs 304.3 GB

12 GB

RTX 3060 12GB · 12 GB tier

DeepSeek V4 Pro

Doesn't fit

smallest quant needs 680 GB

Mistral Large 4 1T A52B

Doesn't fit

smallest quant needs 304.3 GB

16 GB

RTX 5060 Ti 16GB · 16 GB tier

DeepSeek V4 Pro

Doesn't fit

smallest quant needs 680 GB

Mistral Large 4 1T A52B

Doesn't fit

smallest quant needs 304.3 GB

24 GBreference

RTX 4090 · 24 GB tier

DeepSeek V4 Pro

Doesn't fit

smallest quant needs 680 GB

Mistral Large 4 1T A52B

Doesn't fit

smallest quant needs 304.3 GB

32 GB

RTX 5090 · 32 GB tier

DeepSeek V4 Pro

Doesn't fit

smallest quant needs 680 GB

Mistral Large 4 1T A52B

Doesn't fit

smallest quant needs 304.3 GB

48 GB

RTX A6000 · 48 GB tier

DeepSeek V4 Pro

Doesn't fit

smallest quant needs 680 GB

Mistral Large 4 1T A52B

Doesn't fit

smallest quant needs 304.3 GB

Mac 32 GB

Mac Mini M4 · Macs

DeepSeek V4 Pro

Doesn't fit

smallest quant needs 680 GB

Mistral Large 4 1T A52B

Doesn't fit

smallest quant needs 304.3 GB

Mac 64 GB

Mac Studio M4 Max · Macs

DeepSeek V4 Pro

Doesn't fit

smallest quant needs 680 GB

Mistral Large 4 1T A52B

Doesn't fit

smallest quant needs 304.3 GB

Mac 128 GB

Mac Studio M4 Max · Macs

DeepSeek V4 Pro

Doesn't fit

smallest quant needs 680 GB

Mistral Large 4 1T A52B

Doesn't fit

smallest quant needs 304.3 GB

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 Pro Mistral Large 4 1T A52B

Overall

DeepSeek V4 Pro

116.7114–120

Mistral Large 4 1T A52B

118.0115–121

Coding

DeepSeek V4 Pro

110.6105–116

Mistral Large 4 1T A52B

112.3108–117

Agents

DeepSeek V4 Pro

no data

Mistral Large 4 1T A52B

116.6109–125

Math

DeepSeek V4 Pro

117.6113–122

Mistral Large 4 1T A52B

121.7118–125

Science

DeepSeek V4 Pro

122.1110–134

Mistral Large 4 1T A52B

120.4109–132

Reasoning

DeepSeek V4 Pro

118.5112–125

Mistral Large 4 1T A52B

120.3114–127

DeepSeek V4 Pro · #20 of 104 open models · #20 of 104 that fit ≤1 TB

Mistral Large 4 1T A52B · #18 of 104 open models · #16 of 101 that fit ≤640 GB

Benchmarks

18 shared boards: DeepSeek V4 Pro leads 7, Mistral Large 4 1T A52B leads 10 · Benchmarks imported Oct 2026

BoardDeepSeek V4 ProMistral Large 4 1T A52B

GPQA Diamond

DeepSeek V4 Pro

90.9%#6/100 open

IndependentEpoch AI · Apr 2026

Mistral Large 4 1T A52B

88.1%#16/100 open

IndependentEpoch AI · Oct 2026

LiveBench Coding

DeepSeek V4 Pro

70.0#17/20 open

IndependentLiveBench · Jun 2026

Mistral Large 4 1T A52B

77.2#8/20 open

IndependentLiveBench · Jun 2026

LiveBench Reasoning

DeepSeek V4 Pro

82.7#11/20 open

IndependentLiveBench · Jun 2026

Mistral Large 4 1T A52B

83.9#9/20 open

IndependentLiveBench · Jun 2026

AIME 2024/2025

DeepSeek V4 Pro

96.7%#7/88 open

IndependentEpoch AI · Apr 2026

Mistral Large 4 1T A52B

98.3%#4/88 open

IndependentEpoch AI · Oct 2026

FrontierMath

DeepSeek V4 Pro

45.3%#14/19 open

IndependentEpoch AI · Apr 2026

Mistral Large 4 1T A52B

58.6%#8/19 open

IndependentEpoch AI · Oct 2026

LiveBench Math

DeepSeek V4 Pro

90.7#4/20 open

IndependentLiveBench · Jun 2026

Mistral Large 4 1T A52B

93.6#2/20 open

IndependentLiveBench · Jun 2026

SciCode

DeepSeek V4 Pro

50.0%#16/70 open

Lab-reportedEpoch AI · Apr 2026

Mistral Large 4 1T A52B

54.2%#7/70 open

Lab-reportedEpoch AI · Oct 2026

SimpleQA

DeepSeek V4 Pro

47.0%#4/19 open

IndependentEpoch AI · Apr 2026

Mistral Large 4 1T A52B

20.0%#17/19 open

IndependentEpoch AI · Oct 2026

FrontierMath Tier 4

DeepSeek V4 Pro

2.4%#15/15 open

IndependentEpoch AI · Apr 2026

Mistral Large 4 1T A52B

26.8%#8/15 open

IndependentEpoch AI · Oct 2026

FrontierCode

DeepSeek V4 Pro

17.6%#10/13 open

Lab-reportedEpoch AI · Apr 2026

Mistral Large 4 1T A52B

39.4%#4/13 open

Lab-reportedEpoch AI · Oct 2026

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

ALE-Bench

DeepSeek V4 Pro

1006.1#13/46 open

Lab-reportedEpoch AI · Apr 2026

Mistral Large 4 1T A52B

1213.2#8/46 open

Lab-reportedEpoch AI · Oct 2026

Chess Puzzles

DeepSeek V4 Pro

19.0%#14/16 open

IndependentEpoch AI · Apr 2026

Mistral Large 4 1T A52B

33.0%#6/16 open

IndependentEpoch AI · Oct 2026

CritPt

DeepSeek V4 Pro

12.9%#12/82 open

Lab-reportedEpoch AI · Apr 2026

Mistral Large 4 1T A52B

10.6%#14/82 open

Lab-reportedEpoch AI · Oct 2026

DTBench

DeepSeek V4 Pro

90.7%#6/69 open

Lab-reportedEpoch AI · Apr 2026

Mistral Large 4 1T A52B

88.2%#9/69 open

Lab-reportedEpoch AI · Oct 2026

LMCA

DeepSeek V4 Pro

41.2%#8/55 open

Lab-reportedEpoch AI · Apr 2026

Mistral Large 4 1T A52B

35.8%#15/55 open

Lab-reportedEpoch AI · Oct 2026

Mystery Game Puzzles

DeepSeek V4 Pro

17.0%#12/20 open

IndependentEpoch AI · Apr 2026

Mistral Large 4 1T A52B

17.0%#13/20 open

IndependentEpoch AI · Oct 2026

ProofBench

DeepSeek V4 Pro

16.0%#16/29 open

Lab-reportedEpoch AI · Apr 2026

Mistral Large 4 1T A52B

10.0%#20/29 open

Lab-reportedEpoch AI · Oct 2026

Surface Evolver Bench

DeepSeek V4 Pro

40.0%#9/14 open

Lab-reportedEpoch AI · Apr 2026

Mistral Large 4 1T A52B

28.1%#11/14 open

Lab-reportedEpoch AI · Oct 2026

Only DeepSeek V4 Pro (6)

Only Mistral Large 4 1T A52B (10)

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 ProMistral Large 4 1T A52B

Q2_K

DeepSeek V4 Pro

680 GB574.1 GB file

32K 680 GB · 128K 682 GB

Mistral Large 4 1T A52B

429 GB380.7 GB file

32K 483 GB · 128K 657 GB

Q3_K_M

DeepSeek V4 Pro

780 GB748.7 GB file

32K 780 GB · 128K 782 GB

Mistral Large 4 1T A52B

491 GB~487.5 GB file

32K 546 GB · 128K 720 GB

Q4_K_M

DeepSeek V4 Pro

960 GB954.6 GB file

32K 960 GB · 128K 961 GB

Mistral Large 4 1T A52B

604 GB634.8 GB file

32K 658 GB · 128K 832 GB

Q5_K_M

DeepSeek V4 Pro

1140 GB~1139.2 GB file

32K 1140 GB · 128K 1141 GB

Mistral Large 4 1T A52B

716 GB~712.5 GB file

32K 771 GB · 128K 945 GB

Q6_K

DeepSeek V4 Pro

1319 GB~1319.0 GB file

32K 1320 GB · 128K 1321 GB

Mistral Large 4 1T A52B

829 GB~825.0 GB file

32K 883 GB · 128K 1057 GB

Q8_0

DeepSeek V4 Pro

1599 GB1671.8 GB file

32K 1600 GB · 128K 1601 GB

Mistral Large 4 1T A52B

1004 GB~1000.0 GB file

32K 1058 GB · 128K 1232 GB

3 more quants

IQ2_XXS

DeepSeek V4 Pro

not listed

Mistral Large 4 1T A52B

279 GB302.5 GB file

32K 333 GB · 128K 507 GB

IQ3_XXS

DeepSeek V4 Pro

not listed

Mistral Large 4 1T A52B

391 GB423.5 GB file

32K 446 GB · 128K 620 GB

BF16

DeepSeek V4 Pro

3198 GB~3197.7 GB file

32K 3198 GB · 128K 3200 GB

Mistral Large 4 1T A52B

2004 GB~2000.0 GB file

32K 2058 GB · 128K 2232 GB

How to run each

DeepSeek V4 Pro

Ollama · Q4_K_M

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

All ways to run DeepSeek V4 Pro →

Mistral Large 4 1T A52B

Ollama · Q4_K_M

$ollama run hf.co/ggml-org/Mistral-Large-4-1T-A52B-GGUF:Q4_K_M

All ways to run Mistral Large 4 1T A52B →

Specifications

Specifications of DeepSeek V4 Pro and Mistral Large 4 1T A52B
Spec	DeepSeek V4 Pro	Mistral Large 4 1T A52B
Publisher	DeepSeek	Mistral AI
Parameters	1598.8B	1000B
Active parameters	51B per token	52B per token
Architecture	DeepseekV4ForCausalLM	MistralLarge4ForConditionalGeneration
Context	1M tokens	1M tokens
Modalities	Chat	Vision, Function-calling
License	MIT	Other
Released	Apr 2026	Sep 2026

Frequently Asked Questions

Which is better, DeepSeek V4 Pro or Mistral Large 4 1T A52B?

DeepSeek V4 Pro and Mistral Large 4 1T A52B match in llmrun Score, with values of 116.7 versus 118.0; neither fits 24 GB. Choose DeepSeek V4 Pro for also does chat and more permissive license; choose Mistral Large 4 1T A52B for lighter on memory and also does vision, function-calling.

Which needs less VRAM, DeepSeek V4 Pro or Mistral Large 4 1T A52B?

At Q4_K_M, DeepSeek V4 Pro requires 959.6 GB while Mistral Large 4 1T A52B needs 603.9 GB, making Mistral Large 4 1T A52B the easier model to run on your local setup.

Can I run DeepSeek V4 Pro or Mistral Large 4 1T A52B on a 16 GB GPU?

DeepSeek V4 Pro does not fit (smallest quant needs 680 GB); Mistral Large 4 1T A52B does not fit (smallest quant needs 304.3 GB).

Which has a longer context window, DeepSeek V4 Pro or Mistral Large 4 1T A52B?

DeepSeek V4 Pro supports 1,048,576 tokens and Mistral Large 4 1T A52B supports 1,048,576 tokens.