llmrun

Qwen3 235B A22B Instruct 2507 vs Inkling Small

Inkling Small leads llmrun Score 119.6 to 98.5; neither fits 24 GB.

Run Inkling Small in the cloud →

Model A · Alibaba

Qwen3 235B A22B Instruct 2507
98.5llmrun Score

235.1B · 22.2B active · 256K context · Apache 2.0 · Jul 2025

ChatFunctions

Model B · Thinking Machines

Inkling Small
119.6llmrun Score

266.0B · 1.7B active · Apache 2.0 · Jul 2026

VisionChat

Which should you choose?

Choose Qwen3 235B A22B Instruct 2507 if you want…

  • Fits a Mac 128 GB

    Q3_K_L · 122.4 GB; Inkling Small needs 76.3 GB at its smallest quant

  • Also does function-calling

    Inkling Small doesn't list it

Choose Inkling Small if you want…

  • Higher llmrun Score

    119.6 vs 98.5

  • Stronger at reasoning

    122.6 vs 98.1

  • Also does vision

    Qwen3 235B A22B Instruct 2507 doesn't list it

  • Newer release

    Jul 2026 vs Jul 2025

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

Qwen3 235B A22B Instruct 2507

Doesn't fit

smallest quant needs 72.4 GB

Inkling Small

Doesn't fit

smallest quant needs 76.3 GB

12 GB

RTX 3060 12GB · 12 GB tier

Qwen3 235B A22B Instruct 2507

Doesn't fit

smallest quant needs 72.4 GB

Inkling Small

Doesn't fit

smallest quant needs 76.3 GB

16 GB

RTX 5060 Ti 16GB · 16 GB tier

Qwen3 235B A22B Instruct 2507

Doesn't fit

smallest quant needs 72.4 GB

Inkling Small

Doesn't fit

smallest quant needs 76.3 GB

24 GBreference

RTX 4090 · 24 GB tier

Qwen3 235B A22B Instruct 2507

Doesn't fit

smallest quant needs 72.4 GB

Inkling Small

Doesn't fit

smallest quant needs 76.3 GB

32 GB

RTX 5090 · 32 GB tier

Qwen3 235B A22B Instruct 2507

Doesn't fit

smallest quant needs 72.4 GB

Inkling Small

Doesn't fit

smallest quant needs 76.3 GB

48 GB

RTX A6000 · 48 GB tier

Qwen3 235B A22B Instruct 2507

Doesn't fit

smallest quant needs 72.4 GB

Inkling Small

Doesn't fit

smallest quant needs 76.3 GB

Mac 32 GB

Mac Mini M4 · Macs

Qwen3 235B A22B Instruct 2507

Doesn't fit

smallest quant needs 72.4 GB

Inkling Small

Doesn't fit

smallest quant needs 76.3 GB

Mac 64 GB

Mac Studio M4 Max · Macs

Qwen3 235B A22B Instruct 2507

Doesn't fit

smallest quant needs 72.4 GB

Inkling Small

Doesn't fit

smallest quant needs 76.3 GB

Mac 128 GB

Mac Studio M4 Max · Macs

Qwen3 235B A22B Instruct 2507

Q3_K_L · ~21 tok/s

122.4 GB · up to 32Kbetter quant

Inkling Small

Q2_K · ~97 tok/s

116.2 GB · up to 32K

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

Qwen3 235B A22B Instruct 2507

98.595–102

Inkling Small

119.6117–122

Coding

Qwen3 235B A22B Instruct 2507

101.696–107

Inkling Small

no data

Math

Qwen3 235B A22B Instruct 2507

no data

Inkling Small

117.1113–121

Science

Qwen3 235B A22B Instruct 2507

no data

Inkling Small

118.2107–129

Reasoning

Qwen3 235B A22B Instruct 2507

98.193–104

Inkling Small

122.6119–126

Qwen3 235B A22B Instruct 2507 · #63 of 106 open models · #41 of 82 that fit ≤192 GB

Inkling Small · #16 of 106 open models · #6 of 82 that fit ≤192 GB

Benchmarks

2 shared boards: Qwen3 235B A22B Instruct 2507 leads 0, Inkling Small leads 2 · Benchmarks imported Nov 2026

ARC-AGI-2

Qwen3 235B A22B Instruct 2507

1.3%#18/22 open

IndependentEpoch AI · Jul 2025

Inkling Small

40.1%#9/22 open

IndependentEpoch AI · Jul 2026

ARC-AGI

Qwen3 235B A22B Instruct 2507

11.0%#19/22 open

IndependentEpoch AI · Jul 2025

Inkling Small

84.0%#9/22 open

IndependentEpoch AI · Jul 2026

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

Qwen3 235B A22B Instruct 2507

1422.2#33/188 open

ArenaLMArena · Oct 2026

Inkling Small

1405.5#47/188 open

ArenaLMArena · Oct 2026

Only Qwen3 235B A22B Instruct 2507 (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.

Q2_K

Qwen3 235B A22B Instruct 2507

100 GB85.7 GB file

32K 103 GB · 128K 113 GB

Inkling Small

114 GB~113.0 GB file

32K 119 GB · 128K 136 GB

Q3_K_S

Qwen3 235B A22B Instruct 2507

103 GB101.4 GB file

32K 106 GB · 128K 116 GB

Inkling Small

not listed

Q3_K_M

Qwen3 235B A22B Instruct 2507

115 GB112.4 GB file

32K 118 GB · 128K 128 GB

Inkling Small

130 GB119.4 GB file

32K 136 GB · 128K 153 GB

Q4_0

Qwen3 235B A22B Instruct 2507

118 GB133.1 GB file

32K 121 GB · 128K 130 GB

Inkling Small

not listed

Q4_K_M

Qwen3 235B A22B Instruct 2507

142 GB142.2 GB file

32K 145 GB · 128K 154 GB

Inkling Small

160 GB162.5 GB file

32K 166 GB · 128K 182 GB

Q5_K_M

Qwen3 235B A22B Instruct 2507

168 GB166.8 GB file

32K 171 GB · 128K 180 GB

Inkling Small

190 GB196.1 GB file

32K 195 GB · 128K 212 GB

Q6_K

Qwen3 235B A22B Instruct 2507

194 GB193.0 GB file

32K 197 GB · 128K 207 GB

Inkling Small

220 GB218.9 GB file

32K 225 GB · 128K 242 GB

Q8_0

Qwen3 235B A22B Instruct 2507

236 GB249.9 GB file

32K 239 GB · 128K 248 GB

Inkling Small

267 GB280.3 GB file

32K 272 GB · 128K 289 GB

16 more quants

IQ2_XXS

Qwen3 235B A22B Instruct 2507

not listed

Inkling Small

73.8 GB82.3 GB file

32K 79.1 GB · 128K 96.0 GB

IQ2_XS

Qwen3 235B A22B Instruct 2507

71.0 GB65.6 GB file

32K 74.0 GB · 128K 83.5 GB

Inkling Small

not listed

IQ2_S

Qwen3 235B A22B Instruct 2507

74.0 GB66.1 GB file

32K 76.9 GB · 128K 86.4 GB

Inkling Small

not listed

IQ2_M

Qwen3 235B A22B Instruct 2507

79.8 GB78.6 GB file

32K 82.8 GB · 128K 92.3 GB

Inkling Small

90.4 GB82.4 GB file

32K 95.7 GB · 128K 113 GB

IQ3_XXS

Qwen3 235B A22B Instruct 2507

91.6 GB93.1 GB file

32K 94.6 GB · 128K 104 GB

Inkling Small

104 GB97.9 GB file

32K 109 GB · 128K 126 GB

IQ3_XS

Qwen3 235B A22B Instruct 2507

97.5 GB96.9 GB file

32K 100 GB · 128K 110 GB

Inkling Small

not listed

IQ3_S

Qwen3 235B A22B Instruct 2507

not listed

Inkling Small

114 GB107.5 GB file

32K 119 GB · 128K 136 GB

IQ3_M

Qwen3 235B A22B Instruct 2507

106 GB107.3 GB file

32K 109 GB · 128K 119 GB

Inkling Small

not listed

Q3_K_L

Qwen3 235B A22B Instruct 2507

121 GB111.2 GB file

32K 124 GB · 128K 133 GB

Inkling Small

not listed

IQ4_XS

Qwen3 235B A22B Instruct 2507

127 GB125.5 GB file

32K 130 GB · 128K 139 GB

Inkling Small

144 GB127.4 GB file

32K 149 GB · 128K 166 GB

IQ4_NL

Qwen3 235B A22B Instruct 2507

133 GB133.1 GB file

32K 136 GB · 128K 145 GB

Inkling Small

150 GB130.1 GB file

32K 156 GB · 128K 172 GB

Q4_1

Qwen3 235B A22B Instruct 2507

133 GB147.2 GB file

32K 136 GB · 128K 145 GB

Inkling Small

not listed

Q4_K_S

Qwen3 235B A22B Instruct 2507

133 GB133.7 GB file

32K 136 GB · 128K 145 GB

Inkling Small

150 GB152.3 GB file

32K 156 GB · 128K 172 GB

Q4_K_L

Qwen3 235B A22B Instruct 2507

144 GB143.1 GB file

32K 147 GB · 128K 157 GB

Inkling Small

not listed

Q5_K_S

Qwen3 235B A22B Instruct 2507

162 GB161.9 GB file

32K 165 GB · 128K 175 GB

Inkling Small

184 GB184.2 GB file

32K 189 GB · 128K 206 GB

BF16

Qwen3 235B A22B Instruct 2507

471 GB470.3 GB file

32K 474 GB · 128K 483 GB

Inkling Small

533 GB527.4 GB file

32K 538 GB · 128K 555 GB

How to run each

Specifications

Specifications of Qwen3 235B A22B Instruct 2507 and Inkling Small
SpecQwen3 235B A22B Instruct 2507Inkling Small
PublisherAlibabaThinking Machines
Parameters235.1B266.0B
Active parameters22B per token1.7B per token
ArchitectureQwen3MoeForCausalLMInklingForConditionalGeneration
Context256K tokens—
ModalitiesChat, Function-callingVision, Chat
LicenseApache 2.0Apache 2.0
ReleasedJul 2025Jul 2026

Frequently Asked Questions

Which is better, Qwen3 235B A22B Instruct 2507 or Inkling Small?

Inkling Small leads llmrun Score 119.6 to 98.5; neither fits 24 GB. Choose Qwen3 235B A22B Instruct 2507 for fits a Mac 128 GB and also does function-calling; choose Inkling Small for higher llmrun Score and stronger at reasoning.

Which needs less VRAM, Qwen3 235B A22B Instruct 2507 or Inkling Small?

At Q4_K_M, Qwen3 235B A22B Instruct 2507 requires 141.6 GB while Inkling Small needs 160.2 GB, making Qwen3 235B A22B Instruct 2507 the easier model to run on your local setup.

Can I run Qwen3 235B A22B Instruct 2507 or Inkling Small on a 16 GB GPU?

Qwen3 235B A22B Instruct 2507 does not fit (smallest quant needs 72.4 GB); Inkling Small does not fit (smallest quant needs 76.3 GB).