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

144 open-weight models ranked by LMArena score · updated Oct 28, 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	Kimi-K3moonshotai	1476	36.8K	2.8T	Custom	771.9K	11,938
2	GLM-5.3zai-org	1474	26.6K	753B	Custom	2.2M	2,384
3	GLM-5.3-Flashzai-org	1473	41.6K	321B	MIT	6.3M	3,066
4	GLM-5.2zai-org	1470	53.7K	753B	MIT	324.4K	5,202
5	MiMo-V2.5-ProXiaomiMiMo	1464	77.7K	1.0T	MIT	9.9K	771
6	GLM-5.1zai-org	1463	66.8K	754B	MIT	918.3K	1,850
7	Kimi-K2.6moonshotai	1456	40K	1.0T	Modified MIT	238.6K	1,639
8	DeepSeek-V4-Prodeepseek-ai	1451	57.6K	1.6T	MIT	333.6K	5,700
9	GLM-5zai-org	1446	29.3K	754B	MIT	1.5M	2,141
10	NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16nvidia	1445	11.6K	561B	OpenMDW 1.1	1.2M	382
11	Kimi-K2.5moonshotai	1445	74.7K	1.0T	Modified MIT	394.2K	2,898
12	gemma-4-31Bgoogle	1443	6.1K	31.3B	Apache 2.0	61K	589
13	Inklingthinkingmachines	1443	45.5K	952B	Apache 2.0	1.8M	1,837
14	GLM-4.6zai-org	1440	35.8K	357B	MIT	26.9K	1,247
15	Hy3tencent	1439	13.8K	299B	Apache 2.0	1.2M	990
16	Qwen3.8-27BQwen	1438	27.8K	27.8B	Apache 2.0	7.8M	19,243
17	Qwen3.5-397B-A17BQwen	1438	97.7K	403B	Apache 2.0	881.4K	1,581
18	GLM-4.7zai-org	1435	12.2K	358B	MIT	68.4K	2,077
19	gemma-4-26B-A4Bgoogle	1434	6.1K	25.8B	Apache 2.0	72.3K	463
20	DeepSeek-V4-Flashdeepseek-ai	1432	51.9K	291B	MIT	1.1M	2,378
21	MiniMax-M3MiniMaxAI	1431	68.6K	427B	Custom	158.6K	1,601
22	GLM-4.5zai-org	1430	23.8K	358B	MIT	109.4K	1,439
23	Mistral-Large-3-675B-Instruct-2512mistralai	1428	88.3K	—	Apache 2.0	2.5K	261
24	MiMo-V2.5XiaomiMiMo	1428	47.3K	311B	MIT	111.4K	434
25	DeepSeek-R1-0528deepseek-ai	1428	18.2K	685B	MIT	121K	2,494
26	Beam-1.0-501B-A23Breflection	1425	4.2K	501B	Apache 2.0	11.4K	305
27	DeepSeek-V3.2-Expdeepseek-ai	1425	9K	685B	MIT	42.7K	1,027
28	DeepSeek-V3.2deepseek-ai	1424	48.1K	685B	MIT	7.3M	1,594
29	LongCat-Flash-Chatmeituan-longcat	1423	11.2K	562B	MIT	13.5K	540
30	Mistral-Medium-3.5-128Bmistralai	1421	11.8K	128B	Custom	19.4K	497
31	Qwen3-VL-235B-A22B-InstructQwen	1420	11.3K	236B	Apache 2.0	323.5K	430
32	DeepSeek-V3.1deepseek-ai	1420	14.6K	685B	MIT	260K	864
33	Qwen3-235B-A22B-Instruct-2507Qwen	1419	98.1K	235B	Apache 2.0	86.1K	814
34	DeepSeek-V3.1-Terminusdeepseek-ai	1419	3.4K	685B	MIT	9.8K	386
35	Qwen3.5-122B-A10BQwen	1418	29.7K	125B	Apache 2.0	942.5K	653
36	Qwen3-Next-80B-A3B-InstructQwen	1417	22.8K	81.3B	Apache 2.0	129.4K	1,072
37	Qwen3-235B-A22B-Thinking-2507Qwen	1415	8.8K	235B	Apache 2.0	9.1K	419
38	Inkling-Smallthinkingmachines	1414	37K	266B	Apache 2.0	1.3M	443
39	MiMo-V2-FlashXiaomiMiMo	1411	48.1K	310B	MIT	22.8K	756
40	Qwen3.5-27BQwen	1409	28.6K	27.8B	Apache 2.0	1.7M	1,088
41	NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-BF16nvidia	1405	4.2K	124B	OpenMDW 1.1	68.3K	115
42	MiniMax-M2.7MiniMaxAI	1405	89.2K	229B	Custom	314.4K	1,261
43	Step-3.5-Flashstepfun-ai	1403	60.3K	199B	Apache 2.0	220.3K	847
44	Hy3-previewtencent	1401	6.9K	299B	Custom	67K	314
45	Qwen3-VL-235B-A22B-ThinkingQwen	1400	7.9K	236B	Apache 2.0	80.8K	420
46	Qwen3.5-35B-A3BQwen	1395	30.4K	36.0B	Apache 2.0	1.7M	1,554
47	Qwen3-235B-A22BQwen	1393	37.5K	235B	Apache 2.0	295.6K	1,128
48	Muse-Glimmer-30Bmeta-models	1389	3.9K	29.8B	Apache 2.0	185K	2,124
49	Qwen3-30B-A3B-Instruct-2507Qwen	1384	23.3K	30.5B	Apache 2.0	935K	855
50	GLM-4.5-Airzai-org	1384	30.7K	110B	MIT	1.3M	649
51	NVIDIA-Nemotron-3-Super-120B-A12B-BF16nvidia	1377	7.9K	124B	Custom	1.1M	438
52	GLM-4.6Vzai-org	1376	2.8K	108B	MIT	5.4K	398
53	DeepSeek-V3-0324deepseek-ai	1375	45K	685B	MIT	157.7K	3,223
54	DeepSeek-R1deepseek-ai	1373	18.5K	684B	MIT	1.6M	14,434
55	Qwen3-Next-80B-A3B-ThinkingQwen	1368	13.5K	81.3B	Apache 2.0	43.8K	503
56	gpt-oss-120bopenai	1365	30.2K	117B	Apache 2.0	2.4M	5,496
57	granite-4.2-30bibm-granite	1361	3.4K	29.3B	Apache 2.0	30.4K	140
58	Ling-flash-2.0inclusionAI	1361	7K	103B	MIT	4.3K	222
59	MiniMax-M2.5MiniMaxAI	1359	42.8K	229B	Modified MIT	1.3M	1,529
60	gemma-3-27b-itgoogle	1358	46.7K	27.4B	Gemma	361.1K	2,066
61	Qwen3-Coder-480B-A35B-InstructQwen	1357	25.2K	480B	Apache 2.0	57.8K	1,385
62	INTELLECT-3PrimeIntellect	1356	5.4K	107B	MIT	768	218
63	GLM-4.7-Flashzai-org	1351	12K	31.2B	MIT	1M	1,895
64	NVIDIA-Nemotron-3-Nano-30B-A3B-BF16nvidia	1348	15.8K	31.6B	Custom	1.6M	840
65	Trinity-Large-Thinkingarcee-ai	1340	30.3K	399B	OpenMDW 1.1	3.1K	190
66	MiniMax-M2MiniMaxAI	1340	6.9K	229B	Modified MIT	106.3K	1,514
67	Qwen3-32BQwen	1340	3.9K	32.8B	Apache 2.0	2.9M	791
68	Trinity-Large-Previewarcee-ai	1339	31.2K	399B	OpenMDW 1.1	1.6K	181
69	Mistral-Small-3.2-24B-Instruct-2506mistralai	1338	17.5K	24.0B	Apache 2.0	110.3K	644
70	gemma-3-12b-itgoogle	1334	3.8K	12.2B	Gemma	360.2K	907
71	Ring-flash-2.0inclusionAI	1334	7K	103B	MIT	1.3K	104
72	GLM-4.5Vzai-org	1333	4.8K	108B	MIT	16.8K	726
73	DeepSeek-V3deepseek-ai	1332	21.8K	684B	—	1.6M	4,462
74	NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16nvidia	1331	24.2K	31.6B	OpenMDW 1.1	1.1M	269
75	QwQ-32BQwen	1329	25.2K	32.8B	Apache 2.0	32.1K	2,970
76	granite-4.2-8bibm-granite	1319	3.3K	8.8B	Apache 2.0	68.3K	115
77	Qwen3-30B-A3BQwen	1317	26.2K	30.5B	Apache 2.0	1.1M	982
78	Olmo-3.1-32B-Instructallenai	1311	11.8K	32.2B	Apache 2.0	20.4K	88
79	gemma-3n-E4B-itgoogle	1306	22.1K	7.8B	Gemma	12.6K	938
80	Olmo-3-32B-Thinkallenai	1299	5.8K	32.2B	Apache 2.0	12.1K	177
81	DeepSeek-V2.5-1210deepseek-ai	1294	6.8K	236B	Custom	2K	275
82	granite-4.2-3bibm-granite	1293	3.2K	3.7B	Apache 2.0	37.3K	129
83	gemma-3-4b-itgoogle	1291	4.2K	4.3B	Gemma	996.3K	1,629
84	granite-4.1-8bibm-granite	1291	4.2K	8.8B	Apache 2.0	98.3K	257
85	Molmo2-8Ballenai	1287	822	8.7B	Apache 2.0	43.5K	198
86	gpt-oss-20bopenai	1287	10.4K	20.9B	Apache 2.0	5.9M	5,275
87	Llama-4-Maverick-17B-128E-Instructmeta-llama	1287	39.4K	402B	Llama 4	11.6K	520
88	Llama-3.1-405B-Instructmeta-llama	1284	41.4K	406B	Llama 3.1	9.2K	607
89	Llama-3.1-Nemotron-70B-Instructnvidia	1283	7.1K	—	Llama 3.1	72	570
90	Llama-4-Scout-17B-16E-Instructmeta-llama	1279	29.8K	109B	Llama 4	109.2K	1,386
91	Mistral-Small-3.1-24B-Instruct-2503mistralai	1277	32.7K	24.0B	Apache 2.0	169.9K	1,408
92	Llama-3.3-70B-Instructmeta-llama	1274	54.5K	70.6B	Llama 3.3	357.9K	3,193
93	Olmo-3.1-32B-Thinkallenai	1271	8.3K	32.2B	Apache 2.0	6.7K	119
94	DeepSeek-V2.5deepseek-ai	1271	24.6K	236B	Custom	4.5K	750
95	Qwen2.5-72B-InstructQwen	1269	39.4K	72.7B	Custom	185.7K	1,042
96	Llama-3.1-70B-Instructmeta-llama	1261	55.2K	70.6B	Llama 3.1	91.6K	1,027
97	Llama-3.1-Tulu-3-70Ballenai	1256	2.8K	70.6B	Llama 3.1	947	61
98	granite-4.0-h-smallibm-granite	1241	5.6K	32.2B	Apache 2.0	23.3K	313
99	Mistral-Small-24B-Instruct-2501mistralai	1234	14.7K	23.6B	Apache 2.0	94K	991
100	gemma-2-27b-itgoogle	1231	75.8K	27.2B	Gemma	20.5K	577
101	Qwen2.5-Coder-32B-InstructQwen	1230	5.4K	32.8B	Apache 2.0	546K	2,205
102	Nemotron-4-340B-Instructnvidia	1225	19.7K	—	Custom	654	703
103	aya-expanse-32bCohereLabs	1224	27.1K	32.3B	CC BY-NC 4.0	3K	302
104	Meta-Llama-3-70B-Instructmeta-llama	1221	156.9K	70.6B	Llama 3	37.1K	1,537
105	phi-4microsoft	1217	24.1K	14.7B	MIT	299.1K	2,338
106	gemma-2-9b-itgoogle	1207	54.6K	9.2B	Gemma	119K	1,165
107	Qwen2-72B-InstructQwen	1203	37.3K	72.7B	Tongyi Qianwen	24K	724
108	Llama-3.1-Tulu-3-8Ballenai	1193	2.9K	8.0B	Llama 3.1	5K	178
109	DeepSeek-Coder-V2-Instructdeepseek-ai	1191	15.1K	236B	Custom	3.8K	718
110	Llama-3.1-8B-Instructmeta-llama	1187	49.6K	8.0B	Llama 3.1	4.3M	8,816
111	aya-expanse-8bCohereLabs	1185	9.8K	8.0B	CC BY-NC 4.0	24.4K	463
112	Qwen1.5-110B-ChatQwen	1175	26.2K	111B	Tongyi Qianwen	2.7K	139
113	Qwen1.5-72B-ChatQwen	1166	39.3K	72.3B	Tongyi Qianwen	11.1K	224
114	Meta-Llama-3-8B-Instructmeta-llama	1166	104.6K	8.0B	Llama 3	354.8K	5,756
115	Mixtral-8x22B-Instruct-v0.1mistralai	1162	51.4K	141B	Apache 2.0	38.4K	780
116	QwQ-32B-PreviewQwen	1162	3.2K	32.8B	Apache 2.0	7.2K	1,747
117	gemma-2-2b-itgoogle	1156	46.6K	2.6B	Gemma	129.8K	1,605
118	granite-3.1-8b-instructibm-granite	1150	3.1K	8.2B	Apache 2.0	27.9K	169
119	Phi-3-medium-4k-instructmicrosoft	1138	25.1K	14.0B	MIT	18.2K	231
120	Qwen1.5-32B-ChatQwen	1137	21.7K	32.5B	Tongyi Qianwen	46.7K	115
121	Mixtral-8x7B-Instruct-v0.1mistralai	1132	73.5K	46.7B	Apache 2.0	292.7K	4,777
122	Qwen1.5-14B-ChatQwen	1128	17.8K	14.2B	Tongyi Qianwen	12.1K	118
123	granite-3.1-2b-instructibm-granite	1127	3.2K	2.5B	Apache 2.0	24.4K	57
124	Llama-2-70b-chat-hfmeta-llama	1115	38.5K	69.0B	Llama 2	4.6K	2,211
125	Phi-3-small-8k-instructmicrosoft	1110	17.8K	7.4B	MIT	7.1K	182
126	Llama-3.2-3B-Instructmeta-llama	1110	7.9K	3.2B	Llama 3.2	929.3K	2,952
127	deepseek-llm-67b-chatdeepseek-ai	1105	4.9K	—	Custom	2.7K	222
128	granite-3.0-8b-instructibm-granite	1096	6.6K	8.2B	Apache 2.0	13.1K	209
129	gemma-1.1-7b-itgoogle	1094	23.9K	8.5B	Gemma	4.3K	276
130	Mistral-7B-Instruct-v0.2mistralai	1090	19.4K	7.2B	Apache 2.0	1.6M	3,279
131	Llama-2-13b-chat-hfmeta-llama	1084	19.2K	13.0B	Llama 2	152.7K	1,138
132	Qwen1.5-7B-ChatQwen	1083	4.7K	7.7B	Tongyi Qianwen	7.1K	192
133	SOLAR-10.7B-Instruct-v1.0upstage	1083	4.2K	10.7B	CC BY-NC 4.0	48.9K	660
134	Phi-3-mini-4k-instructmicrosoft	1080	12.3K	3.8B	MIT	324.8K	1,489
135	granite-3.0-2b-instructibm-granite	1080	6.8K	2.6B	Apache 2.0	3.2K	48
136	gemma-7b-itgoogle	1056	8.9K	8.5B	Gemma	25.3K	1,263
137	Llama-3.2-1B-Instructmeta-llama	1055	8K	1.2B	Llama 3.2	1.5M	1,958
138	Llama-2-7b-chat-hfmeta-llama	1053	14.1K	6.7B	Llama 2	126.9K	4,949
139	Qwen-14B-ChatQwen	1051	5K	14.2B	—	5.2K	381
140	Phi-3-mini-128k-instructmicrosoft	1050	20.7K	3.8B	MIT	236.2K	1,713
141	OLMo-7B-Instructallenai	1032	6.3K	6.9B	Apache 2.0	845	54
142	gemma-1.1-2b-itgoogle	1022	10.9K	2.5B	Gemma	6.1K	175
143	gemma-2b-itgoogle	1002	4.8K	2.5B	Gemma	13.8K	975
144	Qwen1.5-4B-ChatQwen	997	7.6K	4.0B	Custom	15.9K	53

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.