Best open LLMs

Every open model we can rank, on one 0–100 quality scale fitted across public leaderboards. Filter to the models that fit your GPU.

40 benchmarks4 sourcesUpdated 3 Oct 2026Reference scale 2026-Q4How the score is built

Local models Overall

General Coding Agents Math Science Reasoning

Models ranked by overall llmrun Score
#	Model	Coding	Agents	Math	Science	Reasoning	Benchmarks	VRAM
1	Kimi K31674 GB · 26 benchmarks	73	61	81	52	63	26	1674 GB
2	GLM 5.3456 GB · 21 benchmarks	67	68	72	49	58	21	456 GB
3	DeepSeek V4 Pro 0813991 GB · 21 benchmarks	63	–	71	46	65	21	991 GB
4	DeepSeek V4 Flash 0731183 GB · 21 benchmarks	59	–	71	45	62	21	183 GB
5	DeepSeek V4.1 Flash458 GB · 10 benchmarks	54	–	76	45	57	10	458 GB
6	GLM 5.2456 GB · 24 benchmarks	58	60	68	47	52	24	456 GB
7	Kimi K2.6620 GB · 18 benchmarks	57	–	63	43	51	18	620 GB
8	GLM 5.3 Flash195 GB · 17 benchmarks	52	63	62	44	37	17	195 GB
9	Inkling Small160 GB · 12 benchmarks	–	–	57	41	54	12	160 GB
10	Kimi K2.7 Code620 GB · 19 benchmarks	53	46	61	40	50	19	620 GB
11	GLM 5.1457 GB · 13 benchmarks	53	54	57	40	46	13	457 GB
12	DeepSeek V4 Pro960 GB · 22 benchmarks	49	–	57	44	49	22	960 GB
13	MiMo V2.5 Pro615 GB · 6 benchmarks	54	–	–	38	42	6	615 GB
14	Inkling572 GB · 21 benchmarks	43	–	52	40	53	21	572 GB
15	Qwen3.8 27B17.4 GB · 9 benchmarks	51	–	57	37	50	9	17.4 GB
16	Kimi K2.5620 GB · 20 benchmarks	49	37	–	40	45	20	620 GB
17	Qwen3.5 397B A17B266 GB · 9 benchmarks	–	–	52	–	46	9	266 GB
18	MiniMax M2.5138 GB · 11 benchmarks	47	37	–	–	45	11	138 GB
19	NVIDIA Nemotron 3 Ultra 550B A55B BF16370 GB · 14 benchmarks	43	–	52	37	46	14	370 GB
20	GLM 5457 GB · 15 benchmarks	49	45	–	–	40	15	457 GB
21	Qwen3.6 27B17.4 GB · 12 benchmarks	42	–	53	36	37	12	17.4 GB
22	MiniMax M3257 GB · 19 benchmarks	52	40	46	41	38	19	257 GB
23	DeepSeek V3.2415 GB · 13 benchmarks	35	36	–	–	43	13	415 GB
24	Qwen3.5 122B A10B82.6 GB · 6 benchmarks	–	–	–	26	43	6	82.6 GB
25	DeepSeek V4 Flash175 GB · 10 benchmarks	46	–	–	36	43	10	175 GB
26	DeepSeek V3.2 Exp415 GB · 10 benchmarks	46	–	–	30	44	10	415 GB
27	Kimi K2 Thinking620 GB · 9 benchmarks	43	34	–	–	–	9	620 GB
28	GLM 4.7216 GB · 13 benchmarks	41	32	51	36	35	13	216 GB
29	MiMo V2.5187 GB · 4 benchmarks	43	–	–	33	–	4	187 GB
30	Qwen3 235B A22B Thinking 2507142 GB · 13 benchmarks	42	–	–	34	37	13	142 GB
31	DeepSeek V3.1 Terminus415 GB · 5 benchmarks	47	–	–	31	39	5	415 GB
32	Qwen3.6 35B A3B22.0 GB · 12 benchmarks	44	–	49	35	38	12	22.0 GB
33	MiniMax M2.7138 GB · 6 benchmarks	41	–	–	35	–	6	138 GB
34	Gemma 4 31B IT20.4 GB · 11 benchmarks	47	–	–	32	40	11	20.4 GB
35	Qwen3.5 35B A3B22.0 GB · 8 benchmarks	–	–	–	34	37	8	22.0 GB
36	Gemma 4 26B A4B IT16.1 GB · 9 benchmarks	43	–	–	30	33	9	16.1 GB
37	MiniMax M2138 GB · 5 benchmarks	–	31	–	–	–	5	138 GB
38	DeepSeek R1 0528415 GB · 11 benchmarks	46	–	44	–	32	11	415 GB
39	Qwen3.5 27B17.4 GB · 6 benchmarks	38	–	–	–	41	6	17.4 GB
40	GLM 4.6215 GB · 6 benchmarks	36	31	–	29	–	6	215 GB
41	Seed OSS 36B Instruct22.3 GB · 4 benchmarks	–	–	–	–	–	4	22.3 GB
42	Kimi K2 Instruct 0905620 GB · 6 benchmarks	40	–	–	–	–	6	620 GB
43	GLM 4.5216 GB · 7 benchmarks	38	32	–	–	–	7	216 GB
44	Qwen3 30B A3B Thinking 250718.7 GB · 8 benchmarks	–	–	–	28	27	8	18.7 GB
45	Kimi K2 Instruct620 GB · 10 benchmarks	41	28	–	–	25	10	620 GB
46	DeepSeek V3.1415 GB · 7 benchmarks	–	–	–	–	36	7	415 GB
47	Qwen3.5 9B6.4 GB · 9 benchmarks	–	–	–	32	30	9	6.4 GB
48	DeepSeek R1415 GB · 14 benchmarks	39	–	40	28	29	14	415 GB
49	Qwen3 235B A22B Instruct 2507142 GB · 8 benchmarks	41	–	–	–	31	8	142 GB
50	Qwen3 32B20.3 GB · 9 benchmarks	33	–	–	26	26	9	20.3 GB
51	Qwen3 235B A22B142 GB · 10 benchmarks	41	–	–	–	31	10	142 GB
52	Qwen3 14B9.5 GB · 8 benchmarks	–	–	–	25	24	8	9.5 GB
53	GPT OSS 120B70.5 GB · 20 benchmarks	37	14	–	31	30	20	70.5 GB
54	DeepSeek R1 Distill Qwen 32B20.5 GB · 4 benchmarks	–	–	–	–	–	4	20.5 GB
55	Qwen3 30B A3B Instruct 250718.7 GB · 5 benchmarks	–	–	–	–	26	5	18.7 GB
56	GLM 4.5 Air66.7 GB · 4 benchmarks	–	–	–	–	–	4	66.7 GB
57	GPT OSS 20B12.9 GB · 11 benchmarks	40	–	–	24	26	11	12.9 GB
58	QwQ 32B20.5 GB · 6 benchmarks	–	–	–	–	–	6	20.5 GB
59	NVIDIA Nemotron 3 Super 120B A12B BF1674.7 GB · 4 benchmarks	36	–	–	27	–	4	74.7 GB
60	Qwen3 30B A3B18.7 GB · 7 benchmarks	–	–	–	–	22	7	18.7 GB
61	Qwen3 8B5.5 GB · 8 benchmarks	–	–	–	21	21	8	5.5 GB
62	DeepSeek v3 0324415 GB · 13 benchmarks	39	–	33	27	24	13	415 GB
63	DeepSeek R1 Distill Qwen 14B9.6 GB · 4 benchmarks	–	–	38	–	–	4	9.6 GB
64	Mistral Large 3 675B Instruct 2512446 GB · 5 benchmarks	33	–	–	26	25	5	446 GB
65	Magistral Small 250614.9 GB · 4 benchmarks	–	–	–	–	24	4	14.9 GB
66	DeepSeek v3415 GB · 10 benchmarks	37	–	26	22	21	10	415 GB
67	Llama 4 Maverick 17B 128E Instruct265 GB · 17 benchmarks	25	–	29	26	23	17	265 GB
68	Magistral Small 250915.1 GB · 6 benchmarks	–	–	–	18	22	6	15.1 GB
69	Mistral Small 3.2 24B Instruct 250615.1 GB · 6 benchmarks	–	–	–	18	22	6	15.1 GB
70	Phi 49.5 GB · 6 benchmarks	–	–	26	–	–	6	9.5 GB
71	Qwen2.5 72B Instruct44.6 GB · 9 benchmarks	–	–	25	–	23	9	44.6 GB
72	Gemma 3 27B IT18.1 GB · 11 benchmarks	18	–	29	16	17	11	18.1 GB
73	Llama 4 Scout 17B 16E Instruct71.7 GB · 12 benchmarks	–	–	24	18	20	12	71.7 GB
74	Llama 3.1 405B Instruct268 GB · 8 benchmarks	–	–	22	–	22	8	268 GB
75	Qwen2.5 32B Instruct20.5 GB · 5 benchmarks	–	–	23	–	–	5	20.5 GB
76	Mistral Large Instruct 241174.6 GB · 6 benchmarks	–	–	22	–	22	6	74.6 GB
77	Mistral Large Instruct 240780.9 GB · 7 benchmarks	–	–	20	–	22	7	80.9 GB
78	Llama 3.3 70B Instruct46.6 GB · 13 benchmarks	22	–	19	17	21	13	46.6 GB
79	Mistral Small 3.1 24B Instruct 250315.1 GB · 7 benchmarks	–	–	21	17	21	7	15.1 GB
80	Mistral Small 24B Instruct 250114.9 GB · 4 benchmarks	–	–	20	–	–	4	14.9 GB
81	Llama 3.1 70B Instruct46.6 GB · 8 benchmarks	–	–	18	–	22	8	46.6 GB
82	Qwen2 72B Instruct44.6 GB · 4 benchmarks	–	–	–	–	–	4	44.6 GB
83	Llama 3.2 90B Vision Instruct58.5 GB · 4 benchmarks	–	–	18	–	–	4	58.5 GB
84	Gemma 3 12B IT8.0 GB · 8 benchmarks	–	–	–	12	15	8	8.0 GB
85	Gemma 2 27B IT18.0 GB · 4 benchmarks	–	–	16	–	–	4	18.0 GB
86	Meta Llama 3 70B Instruct46.6 GB · 5 benchmarks	–	–	14	–	–	5	46.6 GB
87	Mixtral 8x22B Instruct v0.185.2 GB · 6 benchmarks	–	–	–	–	19	6	85.2 GB
88	Gemma 2 9B IT6.1 GB · 4 benchmarks	–	–	14	–	–	4	6.1 GB
89	Qwen2.5 7B Instruct5.0 GB · 6 benchmarks	–	–	–	–	15	6	5.0 GB
90	Llama 3.1 8B Instruct5.3 GB · 11 benchmarks	14	–	14	7	16	11	5.3 GB
91	Gemma 3 4B IT2.8 GB · 6 benchmarks	–	–	–	–	16	6	2.8 GB
92	Mistral Nemo Instruct 24078.1 GB · 5 benchmarks	–	–	–	–	–	5	8.1 GB
93	Mixtral 8x7B Instruct v0.128.6 GB · 5 benchmarks	–	–	–	–	16	5	28.6 GB
94	Meta Llama 3 8B Instruct5.3 GB · 7 benchmarks	–	–	10	–	14	7	5.3 GB
95	Llama 3.2 1B Instruct0.8 GB · 4 benchmarks	–	–	–	–	–	4	0.8 GB
96	Deepseek Llm 67B Chat41.3 GB · 4 benchmarks	–	–	10	–	–	4	41.3 GB
97	Mistral 7B Instruct v0.34.9 GB · 4 benchmarks	–	–	10	–	–	4	4.9 GB
98	Llama 2 70B Chat HF45.5 GB · 5 benchmarks	–	–	9	–	10	5	45.5 GB
99	Gemma 3 1B IT0.7 GB · 4 benchmarks	–	–	–	–	–	4	0.7 GB

The range after each score is a 90% interval: given the boards a model has results on, its true score is very likely inside it. Short ranges mean many agreeing results; long ranges mean few or conflicting ones.

Why some models are missing. A model gets a score only with results on at least 4 benchmarks across at least 2 categories, one of them reasoning or coding. Skill scores need at least 2 benchmarks in that skill; otherwise they show as –.

VRAM is llmrun's estimate at Q4_K_M (or the smallest quantization we track) for the weights plus a working context. Proprietary models can't run locally, so the VRAM filter hides them.

How the llmrun Score is built · llmrun does not run these benchmarks; scores are aggregated from public sources.