Best open LLMs for agents

Open models ranked by agentic tool use, from agentic and tool-use boards such as Terminal-Bench and tau-bench. The score shown is the overall llmrun Score; the highlighted column is the agentic tool use skill score used for the order.

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 agents llmrun Score
#	Model	llmrun Score	Coding	Agents	Math	Science	Reasoning	Benchmarks	VRAM
1	GLM 5.3456 GB · 21 benchmarks		67	68	72	49	58	21	456 GB
2	GLM 5.3 Flash195 GB · 17 benchmarks		52	63	62	44	37	17	195 GB
3	Kimi K31674 GB · 26 benchmarks		73	61	81	52	63	26	1674 GB
4	GLM 5.2456 GB · 24 benchmarks		58	60	68	47	52	24	456 GB
5	GLM 5.1457 GB · 13 benchmarks		53	54	57	40	46	13	457 GB
6	Kimi K2.7 Code620 GB · 19 benchmarks		53	46	61	40	50	19	620 GB
7	GLM 5457 GB · 15 benchmarks		49	45	–	–	40	15	457 GB
8	MiniMax M3257 GB · 19 benchmarks		52	40	46	41	38	19	257 GB
9	Kimi K2.5620 GB · 20 benchmarks		49	37	–	40	45	20	620 GB
10	MiniMax M2.5138 GB · 11 benchmarks		47	37	–	–	45	11	138 GB
11	DeepSeek V3.2415 GB · 13 benchmarks		35	36	–	–	43	13	415 GB
12	Kimi K2 Thinking620 GB · 9 benchmarks		43	34	–	–	–	9	620 GB
13	GLM 4.7216 GB · 13 benchmarks		41	32	51	36	35	13	216 GB
14	GLM 4.5216 GB · 7 benchmarks		38	32	–	–	–	7	216 GB
15	Qwen3 Coder 480B A35B Instruct289 GB · 6 benchmarks	—	40	31	–	–	–	6	289 GB
16	MiniMax M2138 GB · 5 benchmarks		–	31	–	–	–	5	138 GB
17	GLM 4.6215 GB · 6 benchmarks		36	31	–	29	–	6	215 GB
18	Kimi K2 Instruct620 GB · 10 benchmarks		41	28	–	–	25	10	620 GB
19	Qwen2.5 Coder 32B Instruct20.5 GB · 3 benchmarks	—	–	15	–	–	–	3	20.5 GB
20	GPT OSS 120B70.5 GB · 20 benchmarks		37	14	–	31	30	20	70.5 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.