llmrun Composite Score

The best open LLMs: every open model we can rank, on one open-ended quality scale fitted across public leaderboards, where 100 is the average model. Filter to the models that fit your GPU.

42 benchmarks4 sourcesUpdated 17 Oct 2026Reference scale 2026-Q4How the Composite Score is built

Composite Score vs VRAM needed

This chart doesn't show proprietary models.

Open models Overall

General Coding Agents Math Science Reasoning

Models ranked by overall llmrun Composite Score
# (open / overall)	Model	Coding	Agents	Math	Science	Reasoning	Benchmarks	VRAM
1 / 31	MiMo V2.6 Pro RL614 GB · 5 benchmarks	124	–	–	140	–	5	614 GB
2 / 45	MiMo V2.6 Flash RL187 GB · 4 benchmarks	124	–	–	122	–	4	187 GB
3 / 47	GLM 5.3454 GB · 23 benchmarks	131	138	129	131	126	23	454 GB
4 / 53	DeepSeek V4.1 Flash458 GB · 20 benchmarks	119	–	129	123	133	20	458 GB
5 / 54	Muse Spark Open 816B A42B552 GB · 5 benchmarks	–	–	128	–	–	5	552 GB
6 / 59	DeepSeek V4 Flash 0731183 GB · 21 benchmarks	121	–	128	124	129	21	183 GB
7 / 69	GLM 5.3 Flash205 GB · 21 benchmarks	113	131	121	124	127	21	205 GB
8 / 74	GLM 5.2454 GB · 26 benchmarks	120	128	126	129	120	26	454 GB
9 / 77	Qwen3.8 27B17.9 GB · 12 benchmarks	112	–	118	112	124	12	17.9 GB
10 / 78	Kimi K2.6618 GB · 20 benchmarks	119	122	122	122	119	20	618 GB
11 / 81	Inkling Small163 GB · 12 benchmarks	–	–	117	118	122	12	163 GB
12 / 84	Kimi K2.7 Code618 GB · 19 benchmarks	115	119	121	118	118	19	618 GB
13 / 87	GLM 5.1454 GB · 14 benchmarks	115	123	117	117	115	14	454 GB
14 / 90	MiMo V2.5 Pro616 GB · 6 benchmarks	116	–	–	115	111	6	616 GB
15 / 91	Inkling575 GB · 22 benchmarks	104	109	112	116	121	22	575 GB
16 / 99	Kimi K2.5618 GB · 20 benchmarks	110	111	–	117	114	20	618 GB
17 / 100	Qwen3.5 397B A17B243 GB · 9 benchmarks	–	–	112	–	115	9	243 GB
18 / 102	MiniMax M2.5140 GB · 11 benchmarks	109	110	–	–	114	11	140 GB
19 / 104	Qwen3.6 27B17.9 GB · 12 benchmarks	102	–	114	111	106	12	17.9 GB
20 / 105	GLM 5454 GB · 15 benchmarks	110	118	–	–	109	15	454 GB
21 / 107	MiniMax M3258 GB · 19 benchmarks	113	113	106	118	107	19	258 GB
22 / 108	NVIDIA Nemotron 3 Ultra 550B A55B BF16378 GB · 15 benchmarks	103	–	112	112	115	15	378 GB
23 / 113	DeepSeek V3.2413 GB · 13 benchmarks	–	108	–	–	112	13	413 GB
24 / 118	Qwen3.5 122B A10B75.5 GB · 6 benchmarks	–	–	–	96	112	6	75.5 GB
25 / 119	DeepSeek V4 Flash175 GB · 10 benchmarks	107	–	–	111	112	10	175 GB
26 / 121	DeepSeek V3.2 Exp413 GB · 10 benchmarks	107	–	–	101	113	10	413 GB
27 / 124	Kimi K2 Thinking617 GB · 9 benchmarks	103	108	–	–	–	9	617 GB
28 / 126	GLM 4.7218 GB · 13 benchmarks	101	106	111	111	104	13	218 GB
29 / 127	MiMo V2.5188 GB · 4 benchmarks	103	–	–	107	–	4	188 GB
30 / 128	Qwen3 235B A22B Thinking 2507143 GB · 13 benchmarks	102	–	–	108	106	13	143 GB
31 / 129	MiniMax M2.7140 GB · 7 benchmarks	103	112	–	112	–	7	140 GB
32 / 130	DeepSeek V3.1 Terminus412 GB · 5 benchmarks	107	–	–	99	108	5	412 GB
33 / 131	Qwen3.6 35B A3B22.0 GB · 12 benchmarks	105	–	109	110	107	12	22.0 GB
34 / 134	Gemma 4 31B IT21.7 GB · 11 benchmarks	108	–	–	105	109	11	21.7 GB
35 / 137	Qwen3.5 35B A3B22.0 GB · 8 benchmarks	–	–	–	108	106	8	22.0 GB
36 / 138	Gemma 4 26B A4B IT16.5 GB · 9 benchmarks	104	–	–	102	101	9	16.5 GB
37 / 147	MiniMax M2140 GB · 5 benchmarks	–	105	–	–	–	5	140 GB
38 / 148	Qwen3.5 27B17.9 GB · 6 benchmarks	98	–	–	–	111	6	17.9 GB
39 / 149	DeepSeek R1 0528412 GB · 10 benchmarks	107	–	–	–	100	10	412 GB
40 / 151	GLM 4.6217 GB · 6 benchmarks	96	104	–	100	–	6	217 GB
41 / 154	Seed OSS 36B Instruct24.1 GB · 4 benchmarks	–	–	–	–	–	4	24.1 GB
42 / 155	Kimi K2 Instruct 0905617 GB · 6 benchmarks	100	–	–	–	–	6	617 GB
43 / 158	GLM 4.5218 GB · 7 benchmarks	99	105	–	–	–	7	218 GB
44 / 159	Qwen3 30B A3B Thinking 250719.4 GB · 8 benchmarks	–	–	–	99	94	8	19.4 GB
45 / 161	Kimi K2 Instruct617 GB · 10 benchmarks	102	102	–	–	90	10	617 GB
46 / 163	DeepSeek V3.1412 GB · 7 benchmarks	–	–	–	–	105	7	412 GB
47 / 165	Qwen3 235B A22B143 GB · 9 benchmarks	101	–	–	–	99	9	143 GB
48 / 166	Qwen3.5 9B6.6 GB · 9 benchmarks	–	–	–	105	97	9	6.6 GB
49 / 167	DeepSeek R1412 GB · 13 benchmarks	100	–	–	99	96	13	412 GB
50 / 168	Qwen3 235B A22B Instruct 2507143 GB · 8 benchmarks	102	–	–	–	98	8	143 GB
51 / 169	Qwen3 32B22.6 GB · 9 benchmarks	93	–	–	96	91	9	22.6 GB
52 / 172	Qwen3 14B11.9 GB · 8 benchmarks	–	–	–	94	88	8	11.9 GB
53 / 173	GPT OSS 120B70.8 GB · 20 benchmarks	98	82	–	103	97	20	70.8 GB
54 / 174	DeepSeek R1 Distill Qwen 32B24.3 GB · 4 benchmarks	–	–	–	–	–	4	24.3 GB
55 / 175	Qwen3 30B A3B Instruct 250719.4 GB · 5 benchmarks	–	–	–	–	92	5	19.4 GB
56 / 176	GLM 4.5 Air67.6 GB · 4 benchmarks	–	–	–	–	–	4	67.6 GB
57 / 177	GPT OSS 20B13.1 GB · 11 benchmarks	101	–	–	92	91	11	13.1 GB
58 / 178	QwQ 32B24.3 GB · 6 benchmarks	–	–	–	–	–	6	24.3 GB
59 / 179	NVIDIA Nemotron 3 Super 120B A12B BF1674.7 GB · 4 benchmarks	96	–	–	98	–	4	74.7 GB
60 / 182	DeepSeek v3 0324412 GB · 12 benchmarks	100	–	–	98	88	12	412 GB
61 / 183	Qwen3 8B7.6 GB · 8 benchmarks	–	–	–	88	81	8	7.6 GB
62 / 184	Qwen3 30B A3B19.4 GB · 7 benchmarks	–	–	–	–	83	7	19.4 GB
63 / 197	Mistral Large 3 675B Instruct 2512454 GB · 5 benchmarks	93	–	–	97	89	5	454 GB
64 / 199	Magistral Small 250617.8 GB · 4 benchmarks	–	–	–	–	88	4	17.8 GB
65 / 200	DeepSeek v3412 GB · 8 benchmarks	97	–	–	90	81	8	412 GB
66 / 202	Llama 4 Maverick 17B 128E Instruct273 GB · 16 benchmarks	81	–	–	96	85	16	273 GB
67 / 203	Mistral Small 3.2 24B Instruct 250618.1 GB · 6 benchmarks	–	–	–	82	83	6	18.1 GB
68 / 204	Magistral Small 250918.1 GB · 6 benchmarks	–	–	–	83	85	6	18.1 GB
69 / 207	Phi 412.5 GB · 5 benchmarks	–	–	–	–	–	5	12.5 GB
70 / 209	Llama 3.1 405B Instruct276 GB · 7 benchmarks	–	–	–	–	85	7	276 GB
71 / 211	Mistral Large Instruct 241179.8 GB · 5 benchmarks	–	–	–	–	84	5	79.8 GB
72 / 212	Llama 4 Scout 17B 16E Instruct77.0 GB · 11 benchmarks	–	–	–	83	80	11	77.0 GB
73 / 213	Qwen2.5 72B Instruct49.3 GB · 8 benchmarks	–	–	–	–	86	8	49.3 GB
74 / 214	Mistral Large Instruct 240786.4 GB · 6 benchmarks	–	–	–	–	83	6	86.4 GB
75 / 216	Qwen2.5 32B Instruct24.3 GB · 4 benchmarks	–	–	–	–	–	4	24.3 GB
76 / 219	Llama 3.3 70B Instruct51.3 GB · 12 benchmarks	76	–	–	81	82	12	51.3 GB
77 / 221	Gemma 3 27B IT21.7 GB · 10 benchmarks	70	–	–	81	74	10	21.7 GB
78 / 222	Mistral Small 3.1 24B Instruct 250318.1 GB · 6 benchmarks	–	–	–	81	82	6	18.1 GB
79 / 227	Llama 3.1 70B Instruct51.3 GB · 7 benchmarks	–	–	–	–	83	7	51.3 GB
80 / 228	Meta Llama 3 70B Instruct51.3 GB · 4 benchmarks	–	–	–	–	–	4	51.3 GB
81 / 229	Gemma 3 12B IT10.9 GB · 8 benchmarks	–	–	–	71	66	8	10.9 GB
82 / 232	Mixtral 8x22B Instruct v0.188.4 GB · 5 benchmarks	–	–	–	–	76	5	88.4 GB
83 / 236	Mistral Nemo Instruct 240711.0 GB · 4 benchmarks	–	–	–	–	–	4	11.0 GB
84 / 238	Mixtral 8x7B Instruct v0.130.5 GB · 4 benchmarks	–	–	–	–	69	4	30.5 GB
85 / 239	Qwen2.5 7B Instruct5.8 GB · 6 benchmarks	–	–	–	–	65	6	5.8 GB
86 / 240	Gemma 3 4B IT5.0 GB · 6 benchmarks	–	–	–	–	69	6	5.0 GB
87 / 241	Llama 3.1 8B Instruct7.9 GB · 10 benchmarks	56	–	–	56	69	10	7.9 GB
88 / 242	Meta Llama 3 8B Instruct7.9 GB · 6 benchmarks	–	–	–	–	64	6	7.9 GB
89 / 243	Llama 3.2 1B Instruct2.3 GB · 4 benchmarks	–	–	–	–	–	4	2.3 GB
90 / 245	Llama 2 70B Chat HF50.2 GB · 4 benchmarks	–	–	–	–	45	4	50.2 GB
91 / 246	Gemma 3 1B IT2.1 GB · 4 benchmarks	–	–	–	–	–	4	2.1 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 receives a score only if it has results on at least 4 benchmarks across at least 2 categories, with one being reasoning or coding. Skill scores require at least 2 benchmarks for that skill; otherwise they display as –.

VRAM is llmrun's estimate at Q4_K_M (or the smallest quantization we track) for the weights plus room for a 16K-token context. Longer contexts need more memory. Proprietary models can't run locally, so the VRAM filter hides them.

How the llmrun Composite Score is built · llmrun doesn't run these benchmarks; scores come from public sources.