Ran it, or it didn’t happen.

Benchmark open models on your own hardware. Runs count once their event log checks out.

Recent verified runsWhereSpeed

PassedFailedNot supported

Cloud leaderboard

Every model runs on the same pinned machine. 17 models, 70 runs.

Suite: [email protected]
Machine: NVIDIA L40S · 48 GB
Engine: ollama 0.40.14

#	Model	Pass rate				Speed	Runs
1	Qwen3.5-35B-A3BQ4_K_M	96%	9/9	8/9	9/9	201.4 tok/s	3
2	Qwen3.5-27BQ4_K_M	93%	8/9	9/9	8/9	35.5 tok/s	3
3	Qwen3-30B-A3B-Instruct-2507Q4_K_M	92%	11/12	10/12	12/12	187.6 tok/s	4
4	GLM-4.7-FlashQ4_K_M	89%	9/9	6/9	9/9	201.7 tok/s	3
5	Qwen3-Coder-30B-A3B-InstructQ4_K_M	89%	9/9	7/9	8/9	186.4 tok/s	3
6	Qwen3-8BQ4_K_M	89%	9/12	11/12	12/12	120.7 tok/s	4
7	Qwen3-14BQ4_K_M	87%	13/15	11/15	15/15	67.1 tok/s	5
8	gemma-3-27b-itQ4_K_M	86%	17/18	Not supported	14/18	36.2 tok/s	6
9	gpt-oss-20bMXFP4	85%	9/9	9/9	5/9	173.8 tok/s	3
10	gemma-3-4b-itQ4_K_M	83%	11/15	Not supported	14/15	232.7 tok/s	5
11	Qwen3.5-2BQ4_K_M	81%	9/12	9/12	11/12	298.1 tok/s	4
12	Qwen3-4BQ4_K_M	78%	6/9	6/9	9/9	247.9 tok/s	3
13	gemma-3-12b-itQ4_K_M	75%	8/12	Not supported	10/12	80.9 tok/s	4
14	Llama-3.1-8B-InstructQ4_K_M	71%	8/15	9/15	15/15	124.4 tok/s	5
15	Phi-3.5-mini-instructQ4_K_M	64%	9/18	Not supported	14/18	261.3 tok/s	6
16	MiniCPM5-2BQ4_K_M	52%	4/9	3/9	7/9	301.4 tok/s	3
17	Mistral-7B-Instruct-v0.2Q4_K_M	47%	6/18	Not supported	11/18	137.1 tok/s	6

Ranked by checks passed. A task a model can’t do is shown, not counted as a zero.

Open model leaderboard

139 open-weight models in math ranked by LMArena score · updated Nov 6, 2026

Top rated

Most downloaded

Newest

All sizes ≤3B 3–9B 9–20B 20–40B 40–80B >80B

#	Model	Score	Votes	Params	License	Downloads	Likes
1	GLM-5.3-Flashzai-org	1502	2.5K	321B	MIT	6.4M	3,088
2	GLM-5.3zai-org	1496	1.3K	753B	Custom	2.1M	2,395
3	Kimi-K3moonshotai	1495	1.8K	2.8T	Custom	771.6K	11,950
4	GLM-5.2zai-org	1485	2.5K	753B	MIT	318.5K	5,205
5	Inklingthinkingmachines	1480	2.3K	952B	Apache 2.0	1.8M	1,842
6	MiMo-V2.5-ProXiaomiMiMo	1480	3.7K	1.0T	MIT	9.9K	771
7	Hy3tencent	1479	791	299B	Apache 2.0	1.2M	990
8	Kimi-K2.6moonshotai	1475	2.1K	1.0T	Modified MIT	237.2K	1,641
9	GLM-5.1zai-org	1473	3.4K	754B	MIT	892.7K	1,850
10	gemma-4-26B-A4Bgoogle	1471	388	25.8B	Apache 2.0	74.1K	465
11	Kimi-K2.5moonshotai	1469	4.2K	1.0T	Modified MIT	395K	2,901
12	Beam-1.0-501B-A23Breflection	1468	238	501B	Apache 2.0	11.8K	311
13	gemma-4-31Bgoogle	1463	431	31.3B	Apache 2.0	57K	589
14	Inkling-Smallthinkingmachines	1458	1.8K	266B	Apache 2.0	1.2M	449
15	NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16nvidia	1457	578	561B	OpenMDW 1.1	1.1M	384
16	Qwen3.8-27BQwen	1455	1.6K	27.8B	Apache 2.0	7.8M	19,348
17	Qwen3.5-397B-A17BQwen	1453	5.5K	403B	Apache 2.0	857.2K	1,581
18	DeepSeek-V4-Prodeepseek-ai	1442	3.1K	1.6T	MIT	335.6K	5,699
19	LongCat-Flash-Chatmeituan-longcat	1442	686	562B	MIT	13.3K	540
20	Qwen3-Next-80B-A3B-InstructQwen	1440	1.2K	81.3B	Apache 2.0	129.9K	1,072
21	GLM-5zai-org	1440	1.7K	754B	MIT	1.4M	2,141
22	MiMo-V2.5XiaomiMiMo	1436	2.5K	311B	MIT	110.3K	434
23	DeepSeek-V3.2deepseek-ai	1435	3.1K	685B	MIT	7.5M	1,596
24	Mistral-Medium-3.5-128Bmistralai	1433	566	128B	Custom	17.2K	498
25	GLM-4.6zai-org	1432	2.2K	357B	MIT	28.1K	1,247
26	Qwen3-235B-A22B-Instruct-2507Qwen	1431	6.1K	235B	Apache 2.0	84.8K	814
27	Qwen3.5-122B-A10BQwen	1431	1.8K	125B	Apache 2.0	924.6K	653
28	Qwen3.5-27BQwen	1429	1.7K	27.8B	Apache 2.0	1.7M	1,090
29	GLM-4.5zai-org	1428	1.4K	358B	MIT	110.4K	1,439
30	DeepSeek-V4-Flashdeepseek-ai	1425	2.7K	291B	MIT	1.1M	2,378
31	Qwen3-VL-235B-A22B-InstructQwen	1425	698	236B	Apache 2.0	327K	431
32	MiniMax-M3MiniMaxAI	1424	3.6K	427B	Custom	155.9K	1,600
33	GLM-4.7zai-org	1424	736	358B	MIT	66.6K	2,077
34	Hy3-previewtencent	1423	431	299B	Custom	68.3K	314
35	DeepSeek-V3.2-Expdeepseek-ai	1422	476	685B	MIT	43.6K	1,027
36	MiniMax-M2.7MiniMaxAI	1422	4.6K	229B	Custom	307.1K	1,261
37	DeepSeek-V3.1deepseek-ai	1420	981	685B	MIT	262.8K	864
38	Mistral-Large-3-675B-Instruct-2512mistralai	1415	5.1K	—	Apache 2.0	2.5K	261
39	Qwen3-235B-A22B-Thinking-2507Qwen	1412	478	235B	Apache 2.0	8.9K	419
40	Muse-Glimmer-30Bmeta-models	1412	219	29.8B	Apache 2.0	187.5K	2,127
41	Qwen3-VL-235B-A22B-ThinkingQwen	1411	423	236B	Apache 2.0	81K	420
42	NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-BF16nvidia	1409	239	124B	OpenMDW 1.1	70.3K	116
43	Step-3.5-Flashstepfun-ai	1408	3.4K	199B	Apache 2.0	219K	847
44	Qwen3.5-35B-A3BQwen	1404	1.8K	36.0B	Apache 2.0	1.6M	1,555
45	DeepSeek-V3.1-Terminusdeepseek-ai	1402	216	685B	MIT	9.5K	386
46	DeepSeek-R1-0528deepseek-ai	1400	861	685B	MIT	122.3K	2,495
47	Qwen3-Next-80B-A3B-ThinkingQwen	1399	824	81.3B	Apache 2.0	44.4K	503
48	Qwen3-32BQwen	1399	316	32.8B	Apache 2.0	3M	791
49	Qwen3-235B-A22BQwen	1397	2.4K	235B	Apache 2.0	308.9K	1,128
50	GLM-4.5-Airzai-org	1396	1.5K	110B	MIT	1.3M	649
51	MiMo-V2-FlashXiaomiMiMo	1395	3K	310B	MIT	22.2K	756
52	Qwen3-30B-A3B-Instruct-2507Qwen	1395	1.4K	30.5B	Apache 2.0	932.4K	855
53	DeepSeek-R1deepseek-ai	1393	1.6K	684B	MIT	1.6M	14,439
54	gpt-oss-120bopenai	1388	1.8K	117B	Apache 2.0	2.4M	5,498
55	NVIDIA-Nemotron-3-Super-120B-A12B-BF16nvidia	1380	547	124B	Custom	1.1M	438
56	MiniMax-M2.5MiniMaxAI	1378	2.5K	229B	Modified MIT	1.2M	1,529
57	INTELLECT-3PrimeIntellect	1376	342	107B	MIT	774	218
58	DeepSeek-V3-0324deepseek-ai	1373	3.2K	685B	MIT	153.8K	3,223
59	NVIDIA-Nemotron-3-Nano-30B-A3B-BF16nvidia	1371	1K	31.6B	Custom	1.6M	840
60	NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16nvidia	1370	1.2K	31.6B	OpenMDW 1.1	1.1M	271
61	Qwen3-Coder-480B-A35B-InstructQwen	1365	1.6K	480B	Apache 2.0	58.6K	1,387
62	Trinity-Large-Thinkingarcee-ai	1365	1.7K	399B	OpenMDW 1.1	3.2K	190
63	Ling-flash-2.0inclusionAI	1364	450	103B	MIT	4.4K	222
64	QwQ-32BQwen	1359	1.7K	32.8B	Apache 2.0	32K	2,971
65	GLM-4.7-Flashzai-org	1355	754	31.2B	MIT	1M	1,896
66	Qwen3-30B-A3BQwen	1354	1.7K	30.5B	Apache 2.0	1.1M	982
67	GLM-4.5Vzai-org	1354	269	108B	MIT	16.2K	726
68	MiniMax-M2MiniMaxAI	1352	317	229B	Modified MIT	109.4K	1,514
69	Ring-flash-2.0inclusionAI	1349	447	103B	MIT	1.4K	104
70	Mistral-Small-3.2-24B-Instruct-2506mistralai	1341	1K	24.0B	Apache 2.0	109.2K	644
71	Trinity-Large-Previewarcee-ai	1335	2K	399B	OpenMDW 1.1	1.7K	181
72	gpt-oss-20bopenai	1317	678	20.9B	Apache 2.0	5.9M	5,286
73	Olmo-3-32B-Thinkallenai	1316	328	32.2B	Apache 2.0	12.2K	177
74	gemma-3-27b-itgoogle	1312	3.6K	27.4B	Gemma	364.5K	2,066
75	granite-4.1-8bibm-granite	1311	248	8.8B	Apache 2.0	98.1K	257
76	DeepSeek-V3deepseek-ai	1311	2.7K	684B	—	1.6M	4,462
77	gemma-3-12b-itgoogle	1307	389	12.2B	Gemma	365.8K	908
78	Olmo-3.1-32B-Instructallenai	1305	693	32.2B	Apache 2.0	20.8K	88
79	Olmo-3.1-32B-Thinkallenai	1304	475	32.2B	Apache 2.0	7K	120
80	Llama-4-Maverick-17B-128E-Instructmeta-llama	1299	2.8K	402B	Llama 4	11.7K	520
81	DeepSeek-V2.5-1210deepseek-ai	1288	1K	236B	Custom	2.1K	275
82	Llama-4-Scout-17B-16E-Instructmeta-llama	1287	1.9K	109B	Llama 4	110.7K	1,387
83	Qwen2.5-72B-InstructQwen	1283	5.4K	72.7B	Custom	192.8K	1,043
84	Llama-3.1-405B-Instructmeta-llama	1281	8.5K	406B	Llama 3.1	9.2K	607
85	Llama-3.1-Nemotron-70B-Instructnvidia	1271	1K	—	Llama 3.1	74	570
86	DeepSeek-V2.5deepseek-ai	1271	3.6K	236B	Custom	4.5K	750
87	Llama-3.3-70B-Instructmeta-llama	1267	5.8K	70.6B	Llama 3.3	357.4K	3,197
88	Mistral-Small-3.1-24B-Instruct-2503mistralai	1261	2.1K	24.0B	Apache 2.0	168.4K	1,408
89	granite-4.0-h-smallibm-granite	1254	343	32.2B	Apache 2.0	23.5K	313
90	Llama-3.1-70B-Instructmeta-llama	1252	7.7K	70.6B	Llama 3.1	92K	1,027
91	gemma-3n-E4B-itgoogle	1251	1.6K	7.8B	Gemma	12.7K	938
92	Qwen2.5-Coder-32B-InstructQwen	1251	725	32.8B	Apache 2.0	548.8K	2,206
93	phi-4microsoft	1246	2.8K	14.7B	MIT	293.7K	2,340
94	Llama-3.1-Tulu-3-70Ballenai	1242	397	70.6B	Llama 3.1	965	61
95	DeepSeek-Coder-V2-Instructdeepseek-ai	1241	1.9K	236B	Custom	3.8K	718
96	Mistral-Small-24B-Instruct-2501mistralai	1240	1.7K	23.6B	Apache 2.0	94.4K	991
97	gemma-3-4b-itgoogle	1239	423	4.3B	Gemma	1M	1,633
98	Qwen2-72B-InstructQwen	1235	4.8K	72.7B	Tongyi Qianwen	24.3K	724
99	Meta-Llama-3-70B-Instructmeta-llama	1218	20.9K	70.6B	Llama 3	37.8K	1,537
100	Nemotron-4-340B-Instructnvidia	1216	2.4K	—	Custom	638	703
101	QwQ-32B-PreviewQwen	1213	480	32.8B	Apache 2.0	7.1K	1,747
102	gemma-2-27b-itgoogle	1212	10.2K	27.2B	Gemma	21.7K	577
103	aya-expanse-32bCohereLabs	1200	3.9K	32.3B	CC BY-NC 4.0	3K	303
104	Llama-3.1-Tulu-3-8Ballenai	1195	363	8.0B	Llama 3.1	5K	178
105	Qwen1.5-110B-ChatQwen	1185	3.2K	111B	Tongyi Qianwen	2.7K	139
106	Mixtral-8x22B-Instruct-v0.1mistralai	1184	6.8K	141B	Apache 2.0	38.7K	780
107	gemma-2-9b-itgoogle	1183	7.1K	9.2B	Gemma	113.6K	1,166
108	Llama-3.1-8B-Instructmeta-llama	1179	7.1K	8.0B	Llama 3.1	4.3M	8,836
109	Phi-3-medium-4k-instructmicrosoft	1173	3.2K	14.0B	MIT	18.7K	231
110	aya-expanse-8bCohereLabs	1168	1.3K	8.0B	CC BY-NC 4.0	22.9K	463
111	Qwen1.5-72B-ChatQwen	1164	5.3K	72.3B	Tongyi Qianwen	11.2K	224
112	granite-3.1-2b-instructibm-granite	1159	391	2.5B	Apache 2.0	24.5K	57
113	Qwen1.5-32B-ChatQwen	1155	2.6K	32.5B	Tongyi Qianwen	46.8K	115
114	Phi-3-mini-4k-instructmicrosoft	1152	1.6K	3.8B	MIT	329.2K	1,489
115	granite-3.1-8b-instructibm-granite	1152	382	8.2B	Apache 2.0	27.4K	169
116	Meta-Llama-3-8B-Instructmeta-llama	1151	14.3K	8.0B	Llama 3	359.6K	5,787
117	Phi-3-small-8k-instructmicrosoft	1151	2.1K	7.4B	MIT	7.1K	182
118	Mixtral-8x7B-Instruct-v0.1mistralai	1147	9.7K	46.7B	Apache 2.0	298.7K	4,777
119	granite-3.0-8b-instructibm-granite	1143	873	8.2B	Apache 2.0	13.3K	209
120	gemma-2-2b-itgoogle	1135	6.6K	2.6B	Gemma	125.7K	1,607
121	Llama-3.2-3B-Instructmeta-llama	1126	1.1K	3.2B	Llama 3.2	914.1K	2,963
122	Qwen1.5-14B-ChatQwen	1125	2.2K	14.2B	Tongyi Qianwen	12.2K	118
123	granite-3.0-2b-instructibm-granite	1117	908	2.6B	Apache 2.0	3.2K	48
124	deepseek-llm-67b-chatdeepseek-ai	1108	576	—	Custom	2.8K	222
125	gemma-1.1-7b-itgoogle	1107	3K	8.5B	Gemma	4.2K	276
126	Llama-2-70b-chat-hfmeta-llama	1091	4.7K	69.0B	Llama 2	4.8K	2,211
127	Phi-3-mini-128k-instructmicrosoft	1089	2.8K	3.8B	MIT	238.7K	1,713
128	Llama-3.2-1B-Instructmeta-llama	1086	1.2K	1.2B	Llama 3.2	1.5M	1,961
129	Mistral-7B-Instruct-v0.2mistralai	1085	2.6K	7.2B	Apache 2.0	1.6M	3,280
130	Qwen1.5-7B-ChatQwen	1080	690	7.7B	Tongyi Qianwen	7K	192
131	Qwen-14B-ChatQwen	1068	534	14.2B	—	5.2K	381
132	gemma-7b-itgoogle	1066	1.1K	8.5B	Gemma	25.3K	1,263
133	Llama-2-13b-chat-hfmeta-llama	1065	2.2K	13.0B	Llama 2	148.8K	1,138
134	SOLAR-10.7B-Instruct-v1.0upstage	1064	604	10.7B	CC BY-NC 4.0	48.7K	660
135	gemma-1.1-2b-itgoogle	1047	1.4K	2.5B	Gemma	6.1K	175
136	Llama-2-7b-chat-hfmeta-llama	1042	1.7K	6.7B	Llama 2	126.4K	4,951
137	Qwen1.5-4B-ChatQwen	1026	988	4.0B	Custom	15.9K	53
138	OLMo-7B-Instructallenai	1018	848	6.9B	Apache 2.0	846	54
139	gemma-2b-itgoogle	1009	597	2.5B	Gemma	14K	975

Show the top 20

Scores: LMArena, CC BY 4.0. Measured on served versions, not local builds. A model with several served variants shows its best.