Mistral Large 4 1T A52B vs Qwen3.8 27B

Qwen3.8 27B leads llmrun Score 121.9 to 118.0 (ranges overlap); only Qwen3.8 27B fits 24 GB (Q4_K_M); Mistral Large 4 1T A52B needs 304.3 GB even at its smallest quant.

Run Mistral Large 4 1T A52B in the cloud →

Model A · Mistral AI

Mistral Large 4 1T A52B

118.0llmrun Score

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

VisionFunctions

Model B · Alibaba

Qwen3.8 27B

121.9llmrun Score

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

VisionFunctions

Which should you choose?

Choose Mistral Large 4 1T A52B if you want…

Stronger at science
120.4 vs 111.9
Longer context window
1M vs 256K tokens

Choose Qwen3.8 27B if you want…

Higher llmrun Score
121.9 vs 118.0, ranges overlap
Stronger at coding
119.0 vs 112.3
Fits a 12 GB GPU
IQ3_XXS · 12 GB; Mistral Large 4 1T A52B needs 304.3 GB at its smallest quant
Fits a Mac 32 GB
Q4_K_M · 17.9 GB; Mistral Large 4 1T A52B needs 304.3 GB at its smallest quant
Lighter on memory
17.4 vs 603.9 GB at Q4_K_M
More permissive license
Apache 2.0 vs Other

On your hardware

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

MemoryMistral Large 4 1T A52BQwen3.8 27B

8 GB

RTX 4060 · 8 GB tier

Mistral Large 4 1T A52B

Doesn't fit

smallest quant needs 304.3 GB

Qwen3.8 27B

Doesn't fit

smallest quant needs 8.8 GB

12 GB

RTX 3060 12GB · 12 GB tier

Mistral Large 4 1T A52B

Doesn't fit

smallest quant needs 304.3 GB

Qwen3.8 27B

IQ3_XXS · ~20 tok/s

12 GB · up to 16Kfits

16 GB

RTX 5060 Ti 16GB · 16 GB tier

Mistral Large 4 1T A52B

Doesn't fit

smallest quant needs 304.3 GB

Qwen3.8 27B

Q3_K_L · ~19 tok/s

15.4 GB · up to 16Kfits

24 GBreference

RTX 4090 · 24 GB tier

Mistral Large 4 1T A52B

Doesn't fit

smallest quant needs 304.3 GB

Qwen3.8 27B

Q4_K_M · ~38 tok/s

17.9 GB · up to 64Kfits

32 GB

RTX 5090 · 32 GB tier

Mistral Large 4 1T A52B

Doesn't fit

smallest quant needs 304.3 GB

Qwen3.8 27B

Q4_K_M · ~67 tok/s

17.9 GB · up to 256Kfits

48 GB

RTX A6000 · 48 GB tier

Mistral Large 4 1T A52B

Doesn't fit

smallest quant needs 304.3 GB

Qwen3.8 27B

Q4_K_M · ~29 tok/s

17.9 GB · up to 256Kfits

Mac 32 GB

Mac Mini M4 · Macs

Mistral Large 4 1T A52B

Doesn't fit

smallest quant needs 304.3 GB

Qwen3.8 27B

Q4_K_M · ~5 tok/s

17.9 GB · up to 128Kfits

Mac 64 GB

Mac Studio M4 Max · Macs

Mistral Large 4 1T A52B

Doesn't fit

smallest quant needs 304.3 GB

Qwen3.8 27B

Q4_K_M · ~22 tok/s

17.9 GB · up to 256Kfits

Mac 128 GB

Mac Studio M4 Max · Macs

Mistral Large 4 1T A52B

Doesn't fit

smallest quant needs 304.3 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

Mistral Large 4 1T A52B Qwen3.8 27B

Overall

Mistral Large 4 1T A52B

118.0115–121

Qwen3.8 27B

121.9120–124

Coding

Mistral Large 4 1T A52B

112.3108–117

Qwen3.8 27B

119.0113–125

Agents

Mistral Large 4 1T A52B

116.6109–125

Qwen3.8 27B

122.6116–129

Math

Mistral Large 4 1T A52B

121.7118–125

Qwen3.8 27B

117.5110–125

Science

Mistral Large 4 1T A52B

120.4109–132

Qwen3.8 27B

111.993–131

Reasoning

Mistral Large 4 1T A52B

120.3114–127

Qwen3.8 27B

124.0121–127

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

Qwen3.8 27B · #13 of 104 open models · #1 of 36 that fit 24 GB

Benchmarks

13 shared boards: Mistral Large 4 1T A52B leads 7, Qwen3.8 27B leads 5 · Benchmarks imported Oct 2026

BoardMistral Large 4 1T A52BQwen3.8 27B

LiveBench Coding

Mistral Large 4 1T A52B

77.2#8/20 open

IndependentLiveBench · Jun 2026

Qwen3.8 27B

75.7#10/20 open

IndependentLiveBench · Jun 2026

LiveBench Reasoning

Mistral Large 4 1T A52B

83.9#9/20 open

IndependentLiveBench · Jun 2026

Qwen3.8 27B

80.0#12/20 open

IndependentLiveBench · Jun 2026

LiveBench Math

Mistral Large 4 1T A52B

93.6#2/20 open

IndependentLiveBench · Jun 2026

Qwen3.8 27B

86.2#12/20 open

IndependentLiveBench · Jun 2026

SciCode

Mistral Large 4 1T A52B

54.2%#7/70 open

Lab-reportedEpoch AI · Oct 2026

Qwen3.8 27B

46.6%#23/70 open

Lab-reportedEpoch AI · Aug 2026

APEX-Agents

Mistral Large 4 1T A52B

51.6%#5/20 open

Lab-reportedEpoch AI · Oct 2026

Qwen3.8 27B

47.5%#8/20 open

Lab-reportedEpoch AI · Aug 2026

CritPt

Mistral Large 4 1T A52B

10.6%#14/82 open

Lab-reportedEpoch AI · Oct 2026

Qwen3.8 27B

5.4%#19/82 open

Lab-reportedEpoch AI · Aug 2026

DTBench

Mistral Large 4 1T A52B

88.2%#9/69 open

Lab-reportedEpoch AI · Oct 2026

Qwen3.8 27B

88.0%#10/69 open

Lab-reportedEpoch AI · Aug 2026

LiveBench 2026 Coding

Mistral Large 4 1T A52B

60.0#14/19 open

IndependentLiveBench · Oct 2026

Qwen3.8 27B

86.8#3/19 open

IndependentLiveBench · Oct 2026

LiveBench 2026 Long-Horizon

Mistral Large 4 1T A52B

11.8#17/19 open

IndependentLiveBench · Oct 2026

Qwen3.8 27B

27.9#10/19 open

IndependentLiveBench · Oct 2026

LiveBench 2026 Reasoning

Mistral Large 4 1T A52B

83.9#6/19 open

IndependentLiveBench · Oct 2026

Qwen3.8 27B

80.0#9/19 open

IndependentLiveBench · Oct 2026

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

LMCA

Mistral Large 4 1T A52B

35.8%#15/55 open

Lab-reportedEpoch AI · Oct 2026

Qwen3.8 27B

41.4%#7/55 open

Lab-reportedEpoch AI · Aug 2026

ProofBench

Mistral Large 4 1T A52B

10.0%#20/29 open

Lab-reportedEpoch AI · Oct 2026

Qwen3.8 27B

16.0%#18/29 open

Lab-reportedEpoch AI · Aug 2026

Surface Evolver Bench

Mistral Large 4 1T A52B

28.1%#11/14 open

Lab-reportedEpoch AI · Oct 2026

Qwen3.8 27B

45.0%#7/14 open

Lab-reportedEpoch AI · Aug 2026

LiveBench 2026 Browser

Mistral Large 4 1T A52B

8.0#13/19 open

IndependentLiveBench · Oct 2026

Qwen3.8 27B

32.0#9/19 open

IndependentLiveBench · Oct 2026

LiveBench 2026 Data Analysis

Mistral Large 4 1T A52B

76.5#6/19 open

IndependentLiveBench · Oct 2026

Qwen3.8 27B

76.6#5/19 open

IndependentLiveBench · Oct 2026

LiveBench 2026 Design

Mistral Large 4 1T A52B

65.6#13/19 open

IndependentLiveBench · Oct 2026

Qwen3.8 27B

75.9#7/19 open

IndependentLiveBench · Oct 2026

LiveBench 2026 Research

Mistral Large 4 1T A52B

28.1#19/19 open

IndependentLiveBench · Oct 2026

Qwen3.8 27B

59.4#15/19 open

IndependentLiveBench · Oct 2026

Only Mistral Large 4 1T A52B (11)

Only Qwen3.8 27B (4)

VRAM by quantization

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

QuantMistral Large 4 1T A52BQwen3.8 27B

Q2_K

Mistral Large 4 1T A52B

429 GB380.7 GB file

32K 483 GB · 128K 657 GB

Qwen3.8 27B

12.6 GB10.8 GB file

32K 13.9 GB · 128K 19.3 GB

Q3_K_S

Mistral Large 4 1T A52B

not listed

Qwen3.8 27B

12.9 GB12.7 GB file

32K 14.2 GB · 128K 19.6 GB

Q3_K_M

Mistral Large 4 1T A52B

491 GB~487.5 GB file

32K 546 GB · 128K 720 GB

Qwen3.8 27B

14.3 GB13.4 GB file

32K 15.6 GB · 128K 21.0 GB

Q4_0

Mistral Large 4 1T A52B

not listed

Qwen3.8 27B

14.6 GB~13.9 GB file

32K 16.0 GB · 128K 21.4 GB

Q4_K_M

Mistral Large 4 1T A52B

604 GB634.8 GB file

32K 658 GB · 128K 832 GB

Qwen3.8 27B

17.4 GB16.5 GB file

32K 18.8 GB · 128K 24.1 GB

Q5_K_M

Mistral Large 4 1T A52B

716 GB~712.5 GB file

32K 771 GB · 128K 945 GB

Qwen3.8 27B

20.5 GB19.8 GB file

32K 21.9 GB · 128K 27.3 GB

Q6_K

Mistral Large 4 1T A52B

829 GB~825.0 GB file

32K 883 GB · 128K 1057 GB

Qwen3.8 27B

23.7 GB22.0 GB file

32K 25.0 GB · 128K 30.4 GB

Q8_0

Mistral Large 4 1T A52B

1004 GB~1000.0 GB file

32K 1058 GB · 128K 1232 GB

Qwen3.8 27B

28.5 GB29.0 GB file

32K 29.9 GB · 128K 35.2 GB

17 more quants

IQ2_XXS

Mistral Large 4 1T A52B

279 GB302.5 GB file

32K 333 GB · 128K 507 GB

Qwen3.8 27B

8.4 GB7.3 GB file

32K 9.7 GB · 128K 15.1 GB

IQ2_XS

Mistral Large 4 1T A52B

not listed

Qwen3.8 27B

9.1 GB9.1 GB file

32K 10.4 GB · 128K 15.8 GB

IQ2_S

Mistral Large 4 1T A52B

not listed

Qwen3.8 27B

9.4 GB8.4 GB file

32K 10.8 GB · 128K 16.1 GB

IQ2_M

Mistral Large 4 1T A52B

not listed

Qwen3.8 27B

10.1 GB10.5 GB file

32K 11.5 GB · 128K 16.8 GB

IQ3_XXS

Mistral Large 4 1T A52B

391 GB423.5 GB file

32K 446 GB · 128K 620 GB

Qwen3.8 27B

11.5 GB10.9 GB file

32K 12.9 GB · 128K 18.2 GB

IQ3_XS

Mistral Large 4 1T A52B

not listed

Qwen3.8 27B

12.2 GB10.4 GB file

32K 13.6 GB · 128K 18.9 GB

IQ3_S

Mistral Large 4 1T A52B

not listed

Qwen3.8 27B

12.6 GB12.0 GB file

32K 13.9 GB · 128K 19.3 GB

IQ3_M

Mistral Large 4 1T A52B

not listed

Qwen3.8 27B

13.3 GB14.9 GB file

32K 14.6 GB · 128K 20.0 GB

Q3_K_L

Mistral Large 4 1T A52B

not listed

Qwen3.8 27B

15.0 GB14.1 GB file

32K 16.3 GB · 128K 21.7 GB

IQ4_XS

Mistral Large 4 1T A52B

not listed

Qwen3.8 27B

15.7 GB14.3 GB file

32K 17.0 GB · 128K 22.4 GB

IQ4_NL

Mistral Large 4 1T A52B

not listed

Qwen3.8 27B

16.4 GB17.4 GB file

32K 17.7 GB · 128K 23.1 GB

Q4_1

Mistral Large 4 1T A52B

not listed

Qwen3.8 27B

16.4 GB17.5 GB file

32K 17.7 GB · 128K 23.1 GB

Q4_K_S

Mistral Large 4 1T A52B

not listed

Qwen3.8 27B

16.4 GB15.4 GB file

32K 17.7 GB · 128K 23.1 GB

Q4_K_L

Mistral Large 4 1T A52B

not listed

Qwen3.8 27B

17.8 GB18.8 GB file

32K 19.1 GB · 128K 24.5 GB

Q5_K_S

Mistral Large 4 1T A52B

not listed

Qwen3.8 27B

19.9 GB18.7 GB file

32K 21.2 GB · 128K 26.6 GB

Q5_K_L

Mistral Large 4 1T A52B

not listed

Qwen3.8 27B

20.9 GB21.5 GB file

32K 22.2 GB · 128K 27.6 GB

BF16

Mistral Large 4 1T A52B

2004 GB~2000.0 GB file

32K 2058 GB · 128K 2232 GB

Qwen3.8 27B

56.3 GB54.7 GB file

32K 57.7 GB · 128K 63.0 GB

How to run each

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 →

Qwen3.8 27B

Ollama · Q4_K_M

$ollama run hf.co/unsloth/Qwen3.8-27B-GGUF:Q4_K_M

All ways to run Qwen3.8 27B →

Specifications

Specifications of Mistral Large 4 1T A52B and Qwen3.8 27B
Spec	Mistral Large 4 1T A52B	Qwen3.8 27B
Publisher	Mistral AI	Alibaba
Parameters	1000B	27.8B
Active parameters	52B per token	—
Architecture	MistralLarge4ForConditionalGeneration	Qwen3_5ForConditionalGeneration
Context	1M tokens	256K tokens
Modalities	Vision, Function-calling	Vision, Function-calling
License	Other	Apache 2.0
Released	Sep 2026	Aug 2026

Frequently Asked Questions

Which is better, Mistral Large 4 1T A52B or Qwen3.8 27B?

Qwen3.8 27B leads llmrun Score 121.9 to 118.0 (ranges overlap); only Qwen3.8 27B fits 24 GB (Q4_K_M); Mistral Large 4 1T A52B needs 304.3 GB even at its smallest quant. Choose Mistral Large 4 1T A52B for stronger at science and longer context window; choose Qwen3.8 27B for higher llmrun Score and stronger at coding.

Which needs less VRAM, Mistral Large 4 1T A52B or Qwen3.8 27B?

At Q4_K_M, Mistral Large 4 1T A52B requires 603.9 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 Mistral Large 4 1T A52B or Qwen3.8 27B on a 16 GB GPU?

Mistral Large 4 1T A52B does not fit (smallest quant needs 304.3 GB); Qwen3.8 27B fits at Q3_K_L.

Which has a longer context window, Mistral Large 4 1T A52B or Qwen3.8 27B?

Mistral Large 4 1T A52B supports 1,048,576 tokens and Qwen3.8 27B supports 262,144 tokens.