MiniMax M3 vs Qwen3.5 397B A17B
MiniMax M3 and Qwen3.5 397B A17B match in llmrun Score, with values of 111.0 versus 112.6; neither fits 24 GB.
Model A · MiniMax
MiniMax M3427.0B · 19.8B active · 1M context · Other · Jun 2026
Model B · Alibaba
Qwen3.5 397B A17B403.4B · 24.4B active · 256K context · Apache 2.0 · Feb 2026
Choose MiniMax M3 if you want…
Longer context window
1M vs 256K tokens
Newer release
Jun 2026 vs Feb 2026
Choose Qwen3.5 397B A17B if you want…
Stronger at reasoning
114.6 vs 107.2
Stronger at math
112.4 vs 106.1
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%.
8 GB
RTX 4060 · 8 GB tier
MiniMax M3
Doesn't fit
smallest quant needs 119.2 GB
Qwen3.5 397B A17B
Doesn't fit
smallest quant needs 111.5 GB
12 GB
RTX 3060 12GB · 12 GB tier
MiniMax M3
Doesn't fit
smallest quant needs 119.2 GB
Qwen3.5 397B A17B
Doesn't fit
smallest quant needs 111.5 GB
16 GB
RTX 5060 Ti 16GB · 16 GB tier
MiniMax M3
Doesn't fit
smallest quant needs 119.2 GB
Qwen3.5 397B A17B
Doesn't fit
smallest quant needs 111.5 GB
24 GBreference
RTX 4090 · 24 GB tier
MiniMax M3
Doesn't fit
smallest quant needs 119.2 GB
Qwen3.5 397B A17B
Doesn't fit
smallest quant needs 111.5 GB
32 GB
RTX 5090 · 32 GB tier
MiniMax M3
Doesn't fit
smallest quant needs 119.2 GB
Qwen3.5 397B A17B
Doesn't fit
smallest quant needs 111.5 GB
48 GB
RTX A6000 · 48 GB tier
MiniMax M3
Doesn't fit
smallest quant needs 119.2 GB
Qwen3.5 397B A17B
Doesn't fit
smallest quant needs 111.5 GB
Mac 32 GB
Mac Mini M4 · Macs
MiniMax M3
Doesn't fit
smallest quant needs 119.2 GB
Qwen3.5 397B A17B
Doesn't fit
smallest quant needs 111.5 GB
Mac 64 GB
Mac Studio M4 Max · Macs
MiniMax M3
Doesn't fit
smallest quant needs 119.2 GB
Qwen3.5 397B A17B
Doesn't fit
smallest quant needs 111.5 GB
Mac 128 GB
Mac Studio M4 Max · Macs
MiniMax M3
IQ2_XXS · ~38 tok/s
119.2 GB · up to 64K
Qwen3.5 397B A17B
IQ2_XS · ~32 tok/s
121.6 GB · up to 128K
Reference GPU: RTX 4090. Tinted cell = the better pick on that machine (fits, more headroom, or at least 10% faster).
Overall
MiniMax M3
Qwen3.5 397B A17B
Coding
MiniMax M3
Qwen3.5 397B A17B
no data
Agents
MiniMax M3
Qwen3.5 397B A17B
no data
Math
MiniMax M3
Qwen3.5 397B A17B
Science
MiniMax M3
Qwen3.5 397B A17B
no data
Reasoning
MiniMax M3
Qwen3.5 397B A17B
MiniMax M3 · #22 of 92 open models · #19 of 88 that fit ≤512 GB
Qwen3.5 397B A17B · #18 of 92 open models · #8 of 77 that fit ≤256 GB
Benchmarks
7 shared boards: MiniMax M3 leads 3, Qwen3.5 397B A17B leads 3 · Benchmarks imported Oct 2026
MiniMax M3
90.9%#6/97 open
IndependentEpoch AI · Jun 2026
Qwen3.5 397B A17B
86.4%#17/97 open
IndependentEpoch AI · Feb 2026
MiniMax M3
1440.5#20/186 open
ArenaLMArena · Oct 2026
Qwen3.5 397B A17B
1441.7#17/186 open
ArenaLMArena · Oct 2026
MiniMax M3
71.1%#26/85 open
IndependentEpoch AI · Jun 2026
Qwen3.5 397B A17B
88.9%#15/85 open
IndependentEpoch AI · Feb 2026
MiniMax M3
1481.7#16/50 open
ArenaLMArena · Oct 2026
Qwen3.5 397B A17B
1399.9#26/50 open
ArenaLMArena · Oct 2026
MiniMax M3
14.0%#16/67 open
IndependentEpoch AI · Jun 2026
Qwen3.5 397B A17B
13.0%#19/67 open
IndependentEpoch AI · Feb 2026
MiniMax M3
78.9%#24/68 open
Lab-reportedEpoch AI · Jun 2026
Qwen3.5 397B A17B
87.5%#12/68 open
Lab-reportedEpoch AI · Feb 2026
MiniMax M3
33.7%#17/54 open
Lab-reportedEpoch AI · Jun 2026
Qwen3.5 397B A17B
37.9%#10/54 open
Lab-reportedEpoch AI · Feb 2026
16 more boards (other shared and single-model boards)
MiniMax M3
37.7%#10/16 open
Lab-reportedEpoch AI · Jun 2026
Qwen3.5 397B A17B
24.9%#13/16 open
Lab-reportedEpoch AI · Feb 2026
MiniMax M3
8.0%#14/18 open
IndependentEpoch AI · Jun 2026
Qwen3.5 397B A17B
18.0%#10/18 open
IndependentEpoch AI · Feb 2026
Only MiniMax M3 (12)
- LiveBench Coding68.2
- LiveBench Reasoning74.5
- LiveBench Math77.0
- SciCode47.1%
- SimpleBench45.8%
- FrontierCode14.7%
- CritPt3.7%
- ForecastBench61.4
- Vending-Bench 22157.8
- ALE-Bench640.0
- Surface Evolver Bench55.0%
- ProofBench18.0%
Only Qwen3.5 397B A17B (2)
- FrontierMath31.2%
- MMLU-Pro87.8%
VRAM by quantization
Memory to load the model and 2K context; below it, the total at 32K and 128K context. Lighter side in bold.
Q2_K
MiniMax M3
182 GB151.9 GB file
32K 185 GB · 128K 194 GB
Qwen3.5 397B A17B
172 GB146.3 GB file
32K 172 GB · 128K 174 GB
Q3_K_S
MiniMax M3
187 GB187.3 GB file
32K 190 GB · 128K 199 GB
Qwen3.5 397B A17B
177 GB164.3 GB file
32K 177 GB · 128K 179 GB
Q3_K_M
MiniMax M3
209 GB195.5 GB file
32K 212 GB · 128K 221 GB
Qwen3.5 397B A17B
197 GB177.4 GB file
32K 197 GB · 128K 199 GB
Q4_0
MiniMax M3
214 GB243.6 GB file
32K 217 GB · 128K 226 GB
Qwen3.5 397B A17B
202 GB~201.7 GB file
32K 203 GB · 128K 204 GB
Q4_K_M
MiniMax M3
257 GB264.9 GB file
32K 260 GB · 128K 269 GB
Qwen3.5 397B A17B
242 GB244.1 GB file
32K 243 GB · 128K 244 GB
Q5_K_M
MiniMax M3
305 GB318.7 GB file
32K 308 GB · 128K 317 GB
Qwen3.5 397B A17B
288 GB293.7 GB file
32K 288 GB · 128K 290 GB
Q6_K
MiniMax M3
353 GB355.3 GB file
32K 356 GB · 128K 365 GB
Qwen3.5 397B A17B
333 GB326.6 GB file
32K 334 GB · 128K 335 GB
Q8_0
MiniMax M3
428 GB453.6 GB file
32K 430 GB · 128K 439 GB
Qwen3.5 397B A17B
404 GB421.5 GB file
32K 404 GB · 128K 406 GB
17 more quants
IQ2_XXS
MiniMax M3
118 GB134.9 GB file
32K 121 GB · 128K 130 GB
Qwen3.5 397B A17B
111 GB114.9 GB file
32K 112 GB · 128K 113 GB
IQ2_XS
MiniMax M3
129 GB129.5 GB file
32K 131 GB · 128K 140 GB
Qwen3.5 397B A17B
121 GB125.4 GB file
32K 122 GB · 128K 123 GB
IQ2_S
MiniMax M3
134 GB132.1 GB file
32K 137 GB · 128K 146 GB
Qwen3.5 397B A17B
126 GB127.3 GB file
32K 127 GB · 128K 128 GB
IQ2_M
MiniMax M3
145 GB135.1 GB file
32K 147 GB · 128K 157 GB
Qwen3.5 397B A17B
137 GB123.1 GB file
32K 137 GB · 128K 138 GB
IQ3_XXS
MiniMax M3
166 GB160.3 GB file
32K 169 GB · 128K 178 GB
Qwen3.5 397B A17B
157 GB140.3 GB file
32K 157 GB · 128K 159 GB
IQ3_XS
MiniMax M3
177 GB196.4 GB file
32K 179 GB · 128K 189 GB
Qwen3.5 397B A17B
167 GB188.2 GB file
32K 167 GB · 128K 169 GB
IQ3_S
MiniMax M3
182 GB175.6 GB file
32K 185 GB · 128K 194 GB
Qwen3.5 397B A17B
172 GB146.4 GB file
32K 172 GB · 128K 174 GB
IQ3_M
MiniMax M3
193 GB205.5 GB file
32K 195 GB · 128K 205 GB
Qwen3.5 397B A17B
182 GB196.0 GB file
32K 182 GB · 128K 184 GB
Q3_K_L
MiniMax M3
219 GB205.0 GB file
32K 222 GB · 128K 231 GB
Qwen3.5 397B A17B
207 GB195.9 GB file
32K 208 GB · 128K 209 GB
IQ4_XS
MiniMax M3
230 GB208.5 GB file
32K 233 GB · 128K 242 GB
Qwen3.5 397B A17B
217 GB189.7 GB file
32K 218 GB · 128K 219 GB
IQ4_NL
MiniMax M3
241 GB212.7 GB file
32K 244 GB · 128K 253 GB
Qwen3.5 397B A17B
227 GB193.8 GB file
32K 228 GB · 128K 229 GB
Q4_1
MiniMax M3
241 GB268.9 GB file
32K 244 GB · 128K 253 GB
Qwen3.5 397B A17B
227 GB256.1 GB file
32K 228 GB · 128K 229 GB
Q4_K_S
MiniMax M3
241 GB248.6 GB file
32K 244 GB · 128K 253 GB
Qwen3.5 397B A17B
227 GB228.0 GB file
32K 228 GB · 128K 229 GB
Q4_K_L
MiniMax M3
not listed
Qwen3.5 397B A17B
248 GB249.0 GB file
32K 248 GB · 128K 249 GB
Q5_K_S
MiniMax M3
294 GB299.7 GB file
32K 297 GB · 128K 306 GB
Qwen3.5 397B A17B
278 GB276.6 GB file
32K 278 GB · 128K 280 GB
Q5_K_L
MiniMax M3
not listed
Qwen3.5 397B A17B
293 GB290.6 GB file
32K 293 GB · 128K 295 GB
BF16
MiniMax M3
855 GB852.9 GB file
32K 857 GB · 128K 866 GB
Qwen3.5 397B A17B
807 GB793.0 GB file
32K 808 GB · 128K 809 GB
MiniMax M3
Ollama · Q4_K_M
ollama run hf.co/unsloth/MiniMax-M3-GGUF:Q4_K_MQwen3.5 397B A17B
Ollama · Q4_K_M
ollama run hf.co/unsloth/Qwen3.5-397B-A17B-GGUF:Q4_K_M| Spec | MiniMax M3 | Qwen3.5 397B A17B |
|---|---|---|
| Publisher | MiniMax | Alibaba |
| Parameters | 427.0B | 403.4B |
| Active parameters | 20B per token | 24B per token |
| Architecture | MiniMaxM3SparseForConditionalGeneration | Qwen3_5MoeForConditionalGeneration |
| Context | 1M tokens | 256K tokens |
| Modalities | Vision, Function-calling | Vision, Function-calling |
| License | Other | Apache 2.0 |
| Released | Jun 2026 | Feb 2026 |
Frequently Asked Questions
Which is better, MiniMax M3 or Qwen3.5 397B A17B?
MiniMax M3 and Qwen3.5 397B A17B match in llmrun Score, with values of 111.0 versus 112.6; neither fits 24 GB. Choose MiniMax M3 for longer context window and newer release; choose Qwen3.5 397B A17B for stronger at reasoning and stronger at math.
Which needs less VRAM, MiniMax M3 or Qwen3.5 397B A17B?
At Q4_K_M, MiniMax M3 requires 256.7 GB while Qwen3.5 397B A17B needs 242.5 GB, making Qwen3.5 397B A17B the easier model to run on your local setup.
Can I run MiniMax M3 or Qwen3.5 397B A17B on a 16 GB GPU?
MiniMax M3 does not fit (smallest quant needs 119.2 GB); Qwen3.5 397B A17B does not fit (smallest quant needs 111.5 GB).
Which has a longer context window, MiniMax M3 or Qwen3.5 397B A17B?
MiniMax M3 supports 1,048,576 tokens and Qwen3.5 397B A17B supports 262,144 tokens.