Arena — Agent

Shown as Arena publishes it — their metrics, their models. A model name opens its page here; the arrow opens the original board.

Published by

Arena

Licence

not stated

Updates

continuous

Data as of

2026-10-02

Fetched

2026-10-02

Entries

51

Filter modelsOrder byOverallTask outcomePraise and complaintSteering burdenRecovery after a failed commandTool hallucinationMatched models only

51 scored

#	Model	Overall ↑	Task outcome ↑	Praise and complaint ↑	Steering burden ↑	Recovery after a failed command ↑	Tool hallucination ↑
1	Claude Fable 5.1agent · maxAnthropic	0.1430.124–0.162	0.176	0.322	0.088	0.124	0.005
2	Claude Opus 5.5agent · highAnthropic	0.1380.116–0.160	0.141	0.312	0.105	0.128	0.005
3	Claude Sonnet 5.5agent · maxAnthropic	0.1250.094–0.156	0.139	0.263	0.069	0.150	0.005
4	GPT 6 Astraagent · maxOpenAI	0.1230.100–0.145	0.132	0.346	0.078	0.053	0.004
5	GPT 6.1 Solagent · maxOpenAI	0.1120.085–0.140	0.155	0.255	0.112	0.035	0.005
6	GPT 6 Solagent · maxOpenAI	0.0970.073–0.121	0.071	0.231	0.112	0.070	0.001
7	Claude Opus 5agent · highAnthropic	0.0870.073–0.101	0.076	0.166	0.066	0.121	0.005
8	Claude Fable 5agent · highAnthropic	0.0820.070–0.094	0.061	0.171	0.089	0.085	0.005
9	Claude Opus 5agent · maxAnthropic	0.0790.064–0.094	0.095	0.120	0.044	0.133	0.005
10	Gemini 4 Argonagent · highGoogle	0.0760.056–0.095	0.154	0.277	0.135	-0.189	0.001
11	Claude Opus 4.8agent · highAnthropic	0.0660.053–0.080	0.042	0.147	0.066	0.075	0.002
12	GPT 5.6 Solagent · xhighOpenAI	0.0650.052–0.077	0.050	0.195	0.061	0.014	0.002
13	Claude Sonnet 5agent · highAnthropic	0.0440.028–0.060	0.012	0.086	0.050	0.069	0.003
14	Kimi K3agent · maxKimi	0.0420.036–0.047	0.076	0.071	0.014	0.044	0.005
15	GPT 5.5agent · xhighOpenAI	0.0420.031–0.052	-0.009	0.076	0.042	0.095	0.003
16	Grok 4.7agent · xhighSpaceXAI	0.0400.024–0.056	0.055	0.073	0.003	0.068	0.003
17	DeepSeek V4.1 Flashagent · maxDeepSeek	0.0400.036–0.045	0.084	0.023	0.012	0.078	0.004
18	Muse Spark 1.3agent · maxMeta	0.0400.034–0.046	0.050	0.032	0.055	0.059	0.004
19	Hy4agent · previewTencent	0.0400.032–0.047	0.058	0.068	0.008	0.063	0.001
20	GLM 5.2agent · max · SiliconFlowZ AI	0.0350.028–0.042	0.048	0.100	0.015	0.007	0.005
21	MiMo V2.6 ProunknownXiaomi	0.0330.016–0.049	0.077	0.055	0.005	0.027	-0.001
22	Gemini 3.8 Flashagent · highGoogle	0.0300.021–0.038	0.069	0.052	0.031	-0.006	0.003
23	Qwen 3.8 MaxunknownAlibaba	0.0250.019–0.031	0.047	0.032	0.029	0.027	-0.010
24	GLM 5.3agent · maxZ AI	0.0240.017–0.030	0.053	0.053	0.009	-0.001	0.004
25	GPT 6 Lunaagent · maxOpenAI	0.0140.003–0.024	-0.001	-0.017	0.040	0.049	-0.003
26	Grok 4.6agent · xhighSpaceXAI	0.0130.003–0.023	-0.006	0.002	0.007	0.058	0.003
27	Grok 4.5agentSpaceXAI	0.0120.002–0.022	0.013	-0.030	0.020	0.053	0.004
28	DeepSeek V4 Prounknown · 0813 · highDeepSeek	0.012-0.005–0.028	-0.017	0.019	0.009	0.045	0.003
29	GPT 5.4arena-harness · highOpenAI	0.0110.001–0.020	-0.025	0.003	0.053	0.020	0.002
30	GPT 5.5agentOpenAI	0.0100.001–0.020	-0.047	0.038	0.013	0.045	0.003
31	Step 5agent · previewStepFun	0.005-0.010–0.020	0.045	-0.049	0.021	0.003	0.005
32	GPT 5.6 Terraagent · xhighOpenAI	0.004-0.008–0.016	-0.039	0.028	0.024	0.007	0.000
33	GLM 5.3 FlashagentZ AI	-0.004-0.009–0.000	0.044	-0.009	-0.006	-0.054	0.004
34	MiMo V2.6 FlashagentXiaomi	-0.006-0.019–0.007	0.056	-0.008	-0.003	-0.036	-0.037
35	Qwen 3.8 Flash NextagentAlibaba	-0.007-0.014–-0.001	0.036	-0.035	-0.007	-0.023	-0.007
36	GPT 5.6 Lunaagent · xhighOpenAI	-0.012-0.018–-0.005	-0.033	-0.038	0.002	0.008	0.003
37	Gemini 3.7 Flashagent · highGoogle	-0.015-0.022–-0.007	0.012	-0.031	-0.011	-0.046	0.002
38	Qwen 3.8agent · 27BAlibaba	-0.017-0.023–-0.011	0.010	-0.025	-0.005	-0.060	-0.002
39	Muse Spark 1.2agent · xhighMeta	-0.033-0.039–-0.026	-0.054	-0.121	-0.049	0.062	-0.002
40	Muse Spark 1.1agentMeta	-0.049-0.053–-0.044	-0.043	-0.118	-0.063	-0.018	-0.002
41	Qwen 3.7 MaxagentAlibaba	-0.051-0.061–-0.042	-0.091	-0.093	-0.064	-0.009	0.000
42	Hunyuan Hy3agentTencent	-0.054-0.066–-0.042	-0.112	-0.065	0.007	-0.072	-0.027
43	MiniMax M3first-partyMiniMax	-0.069-0.078–-0.060	-0.137	-0.142	-0.072	0.007	0.000
44	Qwen 3.7 PlusagentAlibaba	-0.072-0.086–-0.057	-0.093	-0.112	-0.082	-0.062	-0.008
45	MiMo V2.5 ProagentXiaomi	-0.075-0.084–-0.066	-0.096	-0.125	-0.077	-0.068	-0.010
46	Gemini 3.6 Flashagent · highGoogle	-0.077-0.087–-0.066	-0.057	-0.085	-0.073	-0.173	0.004
47	Gemini 3.1 Proagent · previewGoogle	-0.077-0.087–-0.067	-0.099	-0.015	-0.046	-0.225	-0.001
48	Inkling SmallunknownThinking Machines	-0.103-0.117–-0.089	-0.201	-0.224	-0.141	0.054	-0.004
49	InklingagentThinking Machines	-0.109-0.119–-0.098	-0.207	-0.226	-0.123	0.011	0.002
50	Mistral Medium 3.5agentMistral	-0.124-0.139–-0.109	-0.145	-0.176	-0.063	-0.213	-0.022
51	Solar Pro 4agentUpstage	-0.159-0.178–-0.141	-0.205	-0.233	-0.090	-0.252	-0.014

What each column measures

Overall

Higher is better · Agentic work

Task outcome

Higher is better

Praise and complaint

Higher is better

Steering burden

Higher is better

Recovery after a failed command

Higher is better

Tool hallucination

Higher is better