Qwen3.8 27B vs Beam 1.0 501B A23B

Qwen3.8 27B leads llmrun Score 122.0 to 119.5 (ranges overlap); only Qwen3.8 27B fits 24 GB (Q4_K_M); Beam 1.0 501B A23B needs 141.1 GB even at its smallest quant.

Run Beam 1.0 501B A23B in the cloud →

Model A · Alibaba

Qwen3.8 27B

122.0llmrun Score

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

VisionFunctions

Model B · reflection

Beam 1.0 501B A23B

119.5llmrun Score

501.3B · 23.4B active · 1M context · Apache 2.0 · Oct 2026

ChatReasoningCodeFunctions

Which should you choose?

Choose Qwen3.8 27B if you want…

Higher llmrun Score
122.0 vs 119.5, ranges overlap
Fits a 12 GB GPU
IQ3_XXS · 12 GB; Beam 1.0 501B A23B needs 141.1 GB at its smallest quant
Fits a Mac 32 GB
Q4_K_M · 17.9 GB; Beam 1.0 501B A23B needs 141.1 GB at its smallest quant
Lighter on memory
17.4 vs 301.4 GB at Q4_K_M
Also does vision
Beam 1.0 501B A23B doesn't list it

Choose Beam 1.0 501B A23B if you want…

Longer context window
1M vs 256K tokens
Also does chat, reasoning, code
Qwen3.8 27B doesn't list it

On your hardware

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

MemoryQwen3.8 27BBeam 1.0 501B A23B

8 GB

RTX 4060 · 8 GB tier

Qwen3.8 27B

Doesn't fit

smallest quant needs 8.8 GB

Beam 1.0 501B A23B

Doesn't fit

smallest quant needs 141.1 GB

12 GB

RTX 3060 12GB · 12 GB tier

Qwen3.8 27B

IQ3_XXS · ~20 tok/s

12 GB · up to 16Kfits

Beam 1.0 501B A23B

Doesn't fit

smallest quant needs 141.1 GB

16 GB

RTX 5060 Ti 16GB · 16 GB tier

Qwen3.8 27B

Q3_K_L · ~19 tok/s

15.4 GB · up to 16Kfits

Beam 1.0 501B A23B

Doesn't fit

smallest quant needs 141.1 GB

24 GBreference

RTX 4090 · 24 GB tier

Qwen3.8 27B

Q4_K_M · ~38 tok/s

17.9 GB · up to 64Kfits

Beam 1.0 501B A23B

Doesn't fit

smallest quant needs 141.1 GB

32 GB

RTX 5090 · 32 GB tier

Qwen3.8 27B

Q4_K_M · ~67 tok/s

17.9 GB · up to 256Kfits

Beam 1.0 501B A23B

Doesn't fit

smallest quant needs 141.1 GB

48 GB

RTX A6000 · 48 GB tier

Qwen3.8 27B

Q4_K_M · ~29 tok/s

17.9 GB · up to 256Kfits

Beam 1.0 501B A23B

Doesn't fit

smallest quant needs 141.1 GB

Mac 32 GB

Mac Mini M4 · Macs

Qwen3.8 27B

Q4_K_M · ~5 tok/s

17.9 GB · up to 128Kfits

Beam 1.0 501B A23B

Doesn't fit

smallest quant needs 141.1 GB

Mac 64 GB

Mac Studio M4 Max · Macs

Qwen3.8 27B

Q4_K_M · ~22 tok/s

17.9 GB · up to 256Kfits

Beam 1.0 501B A23B

Doesn't fit

smallest quant needs 141.1 GB

Mac 128 GB

Mac Studio M4 Max · Macs

Qwen3.8 27B

Q4_K_M · ~22 tok/s

17.9 GB · up to 256Kfits

Beam 1.0 501B A23B

Doesn't fit

smallest quant needs 141.1 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

Qwen3.8 27B Beam 1.0 501B A23B

Overall

Qwen3.8 27B

122.0120–124

Beam 1.0 501B A23B

119.5117–122

Coding

Qwen3.8 27B

119.0113–125

Beam 1.0 501B A23B

no data

Agents

Qwen3.8 27B

122.7116–129

Beam 1.0 501B A23B

118.3111–126

Math

Qwen3.8 27B

117.5110–125

Beam 1.0 501B A23B

no data

Science

Qwen3.8 27B

111.993–131

Beam 1.0 501B A23B

no data

Reasoning

Qwen3.8 27B

124.0121–127

Beam 1.0 501B A23B

118.8115–122

Qwen3.8 27B · #14 of 106 open models · #1 of 36 that fit 24 GB

Beam 1.0 501B A23B · #17 of 106 open models · #14 of 102 that fit ≤512 GB

Benchmarks

7 shared boards: Qwen3.8 27B leads 4, Beam 1.0 501B A23B leads 2 · Benchmarks imported Nov 2026

BoardQwen3.8 27BBeam 1.0 501B A23B

ARC-AGI-2

Qwen3.8 27B

42.4%#8/22 open

IndependentEpoch AI · Aug 2026

Beam 1.0 501B A23B

23.5%#11/22 open

IndependentEpoch AI · Oct 2026

ARC-AGI

Qwen3.8 27B

87.5%#8/22 open

IndependentEpoch AI · Aug 2026

Beam 1.0 501B A23B

74.0%#12/22 open

IndependentEpoch AI · Oct 2026

APEX-Agents

Qwen3.8 27B

47.5%#8/20 open

Lab-reportedEpoch AI · Aug 2026

Beam 1.0 501B A23B

48.8%#6/20 open

Lab-reportedEpoch AI · Oct 2026

LiveBench 2026 Coding

Qwen3.8 27B

86.8#3/19 open

IndependentLiveBench · Oct 2026

Beam 1.0 501B A23B

83.9#5/19 open

IndependentLiveBench · Oct 2026

LiveBench 2026 Long-Horizon

Qwen3.8 27B

27.9#10/19 open

IndependentLiveBench · Oct 2026

Beam 1.0 501B A23B

17.3#12/19 open

IndependentLiveBench · Oct 2026

LiveBench 2026 Reasoning

Qwen3.8 27B

80.0#9/19 open

IndependentLiveBench · Oct 2026

Beam 1.0 501B A23B

80.2#8/19 open

IndependentLiveBench · Oct 2026

ProofBench

Qwen3.8 27B

16.0%#19/31 open

Lab-reportedEpoch AI · Aug 2026

Beam 1.0 501B A23B

18.0%#16/31 open

Lab-reportedEpoch AI · Oct 2026

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

LMArena Text

Qwen3.8 27B

1435.6#24/188 open

ArenaLMArena · Oct 2026

Beam 1.0 501B A23B

1435.7#23/188 open

ArenaLMArena · Oct 2026

LMArena WebDev

Qwen3.8 27B

1589.8#11/55 open

ArenaLMArena · Oct 2026

Beam 1.0 501B A23B

1436.0#25/55 open

ArenaLMArena · Oct 2026

LiveBench 2026 Browser

Qwen3.8 27B

32.0#9/19 open

IndependentLiveBench · Oct 2026

Beam 1.0 501B A23B

0.0#18/19 open

IndependentLiveBench · Oct 2026

LiveBench 2026 Data Analysis

Qwen3.8 27B

76.6#5/19 open

IndependentLiveBench · Oct 2026

Beam 1.0 501B A23B

66.4#14/19 open

IndependentLiveBench · Oct 2026

LiveBench 2026 Design

Qwen3.8 27B

75.9#7/19 open

IndependentLiveBench · Oct 2026

Beam 1.0 501B A23B

67.9#11/19 open

IndependentLiveBench · Oct 2026

LiveBench 2026 Research

Qwen3.8 27B

59.4#15/19 open

IndependentLiveBench · Oct 2026

Beam 1.0 501B A23B

72.8#8/19 open

IndependentLiveBench · Oct 2026

Only Qwen3.8 27B (8)

VRAM by quantization

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

QuantQwen3.8 27BBeam 1.0 501B A23B

Q2_K

Qwen3.8 27B

12.6 GB10.8 GB file

32K 13.9 GB · 128K 19.3 GB

Beam 1.0 501B A23B

214 GB190.7 GB file

32K 219 GB · 128K 237 GB

Q3_K_S

Qwen3.8 27B

12.9 GB12.7 GB file

32K 14.2 GB · 128K 19.6 GB

Beam 1.0 501B A23B

220 GB225.9 GB file

32K 225 GB · 128K 243 GB

Q3_K_M

Qwen3.8 27B

14.3 GB13.4 GB file

32K 15.6 GB · 128K 21.0 GB

Beam 1.0 501B A23B

245 GB236.6 GB file

32K 250 GB · 128K 268 GB

Q4_0

Qwen3.8 27B

14.6 GB~13.9 GB file

32K 16.0 GB · 128K 21.4 GB

Beam 1.0 501B A23B

251 GB284.2 GB file

32K 257 GB · 128K 274 GB

Q4_K_M

Qwen3.8 27B

17.4 GB16.5 GB file

32K 18.8 GB · 128K 24.1 GB

Beam 1.0 501B A23B

301 GB314.5 GB file

32K 307 GB · 128K 324 GB

Q5_K_M

Qwen3.8 27B

20.5 GB19.8 GB file

32K 21.9 GB · 128K 27.3 GB

Beam 1.0 501B A23B

358 GB380.8 GB file

32K 363 GB · 128K 381 GB

Q6_K

Qwen3.8 27B

23.7 GB22.0 GB file

32K 25.0 GB · 128K 30.4 GB

Beam 1.0 501B A23B

414 GB441.8 GB file

32K 420 GB · 128K 437 GB

Q8_0

Qwen3.8 27B

28.5 GB29.0 GB file

32K 29.9 GB · 128K 35.2 GB

Beam 1.0 501B A23B

502 GB532.8 GB file

32K 507 GB · 128K 525 GB

17 more quants

IQ2_XXS

Qwen3.8 27B

8.4 GB7.3 GB file

32K 9.7 GB · 128K 15.1 GB

Beam 1.0 501B A23B

139 GB145.3 GB file

32K 144 GB · 128K 161 GB

IQ2_XS

Qwen3.8 27B

9.1 GB9.1 GB file

32K 10.4 GB · 128K 15.8 GB

Beam 1.0 501B A23B

151 GB151.7 GB file

32K 156 GB · 128K 174 GB

IQ2_S

Qwen3.8 27B

9.4 GB8.4 GB file

32K 10.8 GB · 128K 16.1 GB

Beam 1.0 501B A23B

157 GB161.3 GB file

32K 163 GB · 128K 180 GB

IQ2_M

Qwen3.8 27B

10.1 GB10.5 GB file

32K 11.5 GB · 128K 16.8 GB

Beam 1.0 501B A23B

170 GB178.5 GB file

32K 175 GB · 128K 193 GB

IQ3_XXS

Qwen3.8 27B

11.5 GB10.9 GB file

32K 12.9 GB · 128K 18.2 GB

Beam 1.0 501B A23B

195 GB211.1 GB file

32K 200 GB · 128K 218 GB

IQ3_XS

Qwen3.8 27B

12.2 GB10.4 GB file

32K 13.6 GB · 128K 18.9 GB

Beam 1.0 501B A23B

207 GB224.5 GB file

32K 213 GB · 128K 230 GB

IQ3_S

Qwen3.8 27B

12.6 GB12.0 GB file

32K 13.9 GB · 128K 19.3 GB

Beam 1.0 501B A23B

not listed

IQ3_M

Qwen3.8 27B

13.3 GB14.9 GB file

32K 14.6 GB · 128K 20.0 GB

Beam 1.0 501B A23B

226 GB255.7 GB file

32K 232 GB · 128K 249 GB

Q3_K_L

Qwen3.8 27B

15.0 GB14.1 GB file

32K 16.3 GB · 128K 21.7 GB

Beam 1.0 501B A23B

258 GB249.5 GB file

32K 263 GB · 128K 280 GB

IQ4_XS

Qwen3.8 27B

15.7 GB14.3 GB file

32K 17.0 GB · 128K 22.4 GB

Beam 1.0 501B A23B

270 GB273.1 GB file

32K 276 GB · 128K 293 GB

IQ4_NL

Qwen3.8 27B

16.4 GB17.4 GB file

32K 17.7 GB · 128K 23.1 GB

Beam 1.0 501B A23B

283 GB311.9 GB file

32K 288 GB · 128K 306 GB

Q4_1

Qwen3.8 27B

16.4 GB17.5 GB file

32K 17.7 GB · 128K 23.1 GB

Beam 1.0 501B A23B

283 GB314.1 GB file

32K 288 GB · 128K 306 GB

Q4_K_S

Qwen3.8 27B

16.4 GB15.4 GB file

32K 17.7 GB · 128K 23.1 GB

Beam 1.0 501B A23B

283 GB293.4 GB file

32K 288 GB · 128K 306 GB

Q4_K_L

Qwen3.8 27B

17.8 GB18.8 GB file

32K 19.1 GB · 128K 24.5 GB

Beam 1.0 501B A23B

not listed

Q5_K_S

Qwen3.8 27B

19.9 GB18.7 GB file

32K 21.2 GB · 128K 26.6 GB

Beam 1.0 501B A23B

345 GB355.1 GB file

32K 351 GB · 128K 368 GB

Q5_K_L

Qwen3.8 27B

20.9 GB21.5 GB file

32K 22.2 GB · 128K 27.6 GB

Beam 1.0 501B A23B

not listed

BF16

Qwen3.8 27B

56.3 GB54.7 GB file

32K 57.7 GB · 128K 63.0 GB

Beam 1.0 501B A23B

1003 GB1002.7 GB file

32K 1009 GB · 128K 1026 GB

How to run each

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 →

Beam 1.0 501B A23B

Ollama · Q4_K_M

$ollama run hf.co/bartowski/Beam-1.0-501B-A23B-GGUF:Q4_K_M

All ways to run Beam 1.0 501B A23B →

Specifications

Specifications of Qwen3.8 27B and Beam 1.0 501B A23B
Spec	Qwen3.8 27B	Beam 1.0 501B A23B
Publisher	Alibaba	reflection
Parameters	27.8B	501.3B
Active parameters	—	23B per token
Architecture	Qwen3_5ForConditionalGeneration	BeamForCausalLM
Context	256K tokens	1M tokens
Modalities	Vision, Function-calling	Chat, Reasoning, Code, Function-calling
License	Apache 2.0	Apache 2.0
Released	Aug 2026	Oct 2026

Frequently Asked Questions

Which is better, Qwen3.8 27B or Beam 1.0 501B A23B?

Qwen3.8 27B leads llmrun Score 122.0 to 119.5 (ranges overlap); only Qwen3.8 27B fits 24 GB (Q4_K_M); Beam 1.0 501B A23B needs 141.1 GB even at its smallest quant. Choose Qwen3.8 27B for higher llmrun Score and fits a 12 GB GPU; choose Beam 1.0 501B A23B for longer context window and also does chat, reasoning, code.

Which needs less VRAM, Qwen3.8 27B or Beam 1.0 501B A23B?

At Q4_K_M, Qwen3.8 27B requires 17.4 GB while Beam 1.0 501B A23B needs 301.4 GB, making Qwen3.8 27B the easier model to run on your local setup.

Can I run Qwen3.8 27B or Beam 1.0 501B A23B on a 16 GB GPU?

Qwen3.8 27B fits at Q3_K_L; Beam 1.0 501B A23B does not fit (smallest quant needs 141.1 GB).

Which has a longer context window, Qwen3.8 27B or Beam 1.0 501B A23B?

Qwen3.8 27B supports 262,144 tokens and Beam 1.0 501B A23B supports 1,048,576 tokens.