The leaders, at a glance

Agentic AI model ranking

Compare AI models and effort settings for agentic work using sourced task evaluations and visible evidence coverage.

See the full ranking ↓How configurations are ranked

Agentic leaders

Experimental comparison score · 0–100

For the selected profile and sources. Scores are experimental; small gaps may not be meaningful. How scoring works · Effort evidence collected 2026-09-12

451unique modelsEach model counted once across effort levels 492model configurationsDocumented model + effort settings, including unranked

109aggregated sourcesOriginal reports and benchmark leaderboards

Explore all 109 sources Independent evaluations and lab reports

Distinct source pages with published results in the catalog. Multiple results and page sections count as one source.

Every ranked row is a measured configuration. The default view shows the highest-scoring supported configuration per model. In Value, the budget is applied first. Select All efforts for the full comparison. How effort ranking works.

Customize ranking

Full ranking

127 entries

Custom capability ranking of measured model configurations. Scores are experimental; missing results remain unknown.
Rank	Model	Score	Capability coverage	analyst-agent	briefcase	aime	apex-agents	aa-lcr	automationbench	critpt	enterprise-ops	gdp-pdf	gdpval	gmmlu	gpqa	harvey	hle	ifbench	itbench	livecodebench	math500	mmlu-pro	mmmu-pro	omniscience	scicode	tau-bench	terminal-bench	arc-agi	deepswe	swe-bench-pro	arc-agi-3	frontiercode	enigma-eval	swe-atlas-qna	swe-atlas-refactoring	swe-atlas-test-writing	livebench-math	livebench-coding	livebench-language	livebench-data	multi-swe-bench	livebench-reasoning	livebench-instructions	swe-rebench	epoch-game-puzzles	frontiermath	simpleqa	vals-finance-agent	vals-code-migration	vals-legal-research	vals-excel-modeling	vibe-code	chartography	osworld	charxiv	swe-bench-multilingual	deepsearchqa	officeqa	mcp-atlas	toolathlon	mrcr	browsecomp
26	Muse Spark 1.1 · XHighMeta Measured effort: XHigh	50.4	Eligible · 100%49 published · 10 families · 1 selected capabilities	—	55.687/87 peersIndependent evidence	—	—	83.3197/197 peersIndependent evidence	73.989/89 peersIndependent evidence	89.1200/200 peersIndependent evidence	—	75.288/88 peersIndependent evidence	78.6123/123 peersIndependent evidence	—	86.9212/212 peersIndependent evidence	92.732/32 peersIndependent evidence	96.4208/208 peersIndependent evidence	—	—	—	—	62.731/31 peersIndependent evidence	—	95.1198/198 peersIndependent evidence	96.391/91 peersIndependent evidence	80.3112/112 peersIndependent evidence	59.8126/126 peersIndependent evidence	—	37.625/25 peersIndependent evidence	—	—	—	—	—	—	100.01/1 peersIndependent evidence	31.426/26 peersIndependent evidence	37.726/26 peersIndependent evidence	26.226/26 peersIndependent evidence	49.126/26 peersIndependent evidence	76.326/26 peersIndependent evidence	53.026/26 peersIndependent evidence	55.026/26 peersIndependent evidence	—	—	—	—	76.830/30 peersIndependent evidence	39.733/33 peersIndependent evidence	43.232/32 peersIndependent evidence	37.632/32 peersIndependent evidence	55.232/32 peersIndependent evidence	—	—	—	—	—	—	—	—	—	—
27	Grok-4.5 · HighxAI Measured effort: High	49.7	Eligible · 100%42 published · 11 families · 1 selected capabilities	52.128/28 peersIndependent evidence	82.887/87 peersIndependent evidence	82.848/48 peersIndependent evidence	—	71.9198/198 peersIndependent evidence	92.789/89 peersIndependent evidence	89.2200/200 peersIndependent evidence	55.041/41 peersIndependent evidence	85.188/88 peersIndependent evidence	85.4123/123 peersIndependent evidence	—	89.9213/213 peersIndependent evidence	87.250/50 peersIndependent evidence	93.6208/208 peersIndependent evidence	—	—	—	—	64.431/31 peersIndependent evidence	90.0114/114 peersIndependent evidence	94.4198/198 peersIndependent evidence	85.991/91 peersIndependent evidence	91.0112/112 peersIndependent evidence	7.8126/126 peersIndependent evidence	54.638/38 peersIndependent evidence	46.525/25 peersIndependent evidence	—	28.86/6 peersIndependent evidence	33.31/1 peersIndependent evidence	—	—	—	—	—	—	—	—	—	—	—	57.17/7 peersIndependent evidence	79.950/50 peersIndependent evidence	33.939/39 peersIndependent evidence	44.025/25 peersIndependent evidence	23.430/30 peersIndependent evidence	48.633/33 peersIndependent evidence	42.332/32 peersIndependent evidence	25.132/32 peersIndependent evidence	40.732/32 peersIndependent evidence	—	—	—	—	—	—	—	—	—	—
28	DeepSeek-V4-Flash-0731 · MaxDeepSeek Measured effort: Max	48.2	Eligible · 100%24 published · 5 families · 1 selected capabilities	—	81.387/87 peersIndependent evidence	74.048/48 peersIndependent evidence	—	74.9198/198 peersIndependent evidence	84.889/89 peersIndependent evidence	90.8200/200 peersIndependent evidence	—	61.888/88 peersIndependent evidence	88.0123/123 peersIndependent evidence	—	83.2213/213 peersIndependent evidence	—	88.7208/208 peersIndependent evidence	—	—	—	—	—	—	87.2198/198 peersIndependent evidence	69.591/91 peersIndependent evidence	87.7112/112 peersIndependent evidence	83.5125/125 peersIndependent evidence	74.338/38 peersIndependent evidence	36.925/25 peersIndependent evidence	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	75.860/60 peersIndependent evidence	39.739/39 peersIndependent evidence	21.225/25 peersIndependent evidence	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—
29	Gemini 3.5 Flash · HighGoogle Measured effort: High	44.9	Eligible · 100%65 published · 14 families · 1 selected capabilities	71.628/28 peersIndependent evidence	57.687/87 peersIndependent evidence	—	100.026/26 peersIndependent evidence	71.4198/198 peersIndependent evidence	75.089/89 peersIndependent evidence	88.4200/200 peersIndependent evidence	95.141/41 peersIndependent evidence	86.488/88 peersIndependent evidence	75.9123/123 peersIndependent evidence	—	92.9212/212 peersIndependent evidence	35.450/50 peersIndependent evidence	94.4208/208 peersIndependent evidence	94.6178/178 peersIndependent evidence	76.228/28 peersIndependent evidence	—	—	74.031/31 peersIndependent evidence	97.4114/114 peersIndependent evidence	94.5198/198 peersIndependent evidence	80.991/91 peersIndependent evidence	85.7202/202 peersIndependent evidence	70.7201/201 peersIndependent evidence	78.738/38 peersIndependent evidence	13.125/25 peersIndependent evidence	—	—	—	15.34/4 peersIndependent evidence	—	—	—	45.026/26 peersIndependent evidence	41.426/26 peersIndependent evidence	72.526/26 peersIndependent evidence	32.626/26 peersIndependent evidence	35.526/26 peersIndependent evidence	29.826/26 peersIndependent evidence	72.126/26 peersIndependent evidence	—	83.140/40 peersIndependent evidence	46.839/39 peersIndependent evidence	82.925/25 peersIndependent evidence	83.430/30 peersIndependent evidence	27.933/33 peersIndependent evidence	28.832/32 peersIndependent evidence	53.932/32 peersIndependent evidence	25.632/32 peersIndependent evidence	—	—	—	—	—	—	—	—	—	—
30	Gemini 3.6 Flash · HighGoogle Measured effort: High	41.6	Eligible · 100%59 published · 10 families · 1 selected capabilities	—	65.387/87 peersIndependent evidence	75.548/48 peersIndependent evidence	—	72.8198/198 peersIndependent evidence	81.989/89 peersIndependent evidence	84.9200/200 peersIndependent evidence	—	81.688/88 peersIndependent evidence	81.3123/123 peersIndependent evidence	—	96.1213/213 peersIndependent evidence	46.050/50 peersIndependent evidence	90.1208/208 peersIndependent evidence	—	—	—	—	67.431/31 peersIndependent evidence	96.3114/114 peersIndependent evidence	93.6198/198 peersIndependent evidence	80.291/91 peersIndependent evidence	77.0112/112 peersIndependent evidence	68.3126/126 peersIndependent evidence	66.538/38 peersIndependent evidence	25.525/25 peersIndependent evidence	—	—	52.06/6 peersIndependent evidence	—	—	—	—	28.326/26 peersIndependent evidence	40.826/26 peersIndependent evidence	69.026/26 peersIndependent evidence	29.026/26 peersIndependent evidence	20.426/26 peersIndependent evidence	45.426/26 peersIndependent evidence	76.826/26 peersIndependent evidence	—	72.560/60 peersIndependent evidence	33.639/39 peersIndependent evidence	78.725/25 peersIndependent evidence	69.730/30 peersIndependent evidence	34.133/33 peersIndependent evidence	22.432/32 peersIndependent evidence	62.932/32 peersIndependent evidence	35.132/32 peersIndependent evidence	—	—	—	—	—	—	—	—	—	—
31	GLM-5.2 · MaxZ.ai Measured effort: Max	41.4	Eligible · 100%40 published · 13 families · 1 selected capabilities	—	78.487/87 peersIndependent evidence	71.148/48 peersIndependent evidence	88.726/26 peersIndependent evidence	61.5198/198 peersIndependent evidence	22.189/89 peersIndependent evidence	95.5200/200 peersIndependent evidence	66.641/41 peersIndependent evidence	57.688/88 peersIndependent evidence	86.2123/123 peersIndependent evidence	—	65.7213/213 peersIndependent evidence	72.150/50 peersIndependent evidence	92.3208/208 peersIndependent evidence	89.2178/178 peersIndependent evidence	87.728/28 peersIndependent evidence	—	—	—	—	64.3198/198 peersIndependent evidence	74.191/91 peersIndependent evidence	91.6202/202 peersIndependent evidence	54.0201/201 peersIndependent evidence	—	15.325/25 peersIndependent evidence	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	73.050/50 peersIndependent evidence	57.539/39 peersIndependent evidence	28.225/25 peersIndependent evidence	—	58.833/33 peersIndependent evidence	41.132/32 peersIndependent evidence	—	39.232/32 peersIndependent evidence	—	—	—	—	—	0.04/4 peersPublisher reports	0.04/4 peersPublisher reports	0.04/4 peersPublisher reports	—	—
32	Claude Sonnet 4.6 · MaxAnthropic Measured effort: Max	40.8	Eligible · 100%39 published · 12 families · 1 selected capabilities	45.828/28 peersIndependent evidence	70.687/87 peersIndependent evidence	47.348/48 peersIndependent evidence	61.226/26 peersIndependent evidence	85.0198/198 peersIndependent evidence	57.389/89 peersIndependent evidence	75.5200/200 peersIndependent evidence	—	78.488/88 peersIndependent evidence	78.1123/123 peersIndependent evidence	91.563/63 peersIndependent evidence	47.9213/213 peersIndependent evidence	51.950/50 peersIndependent evidence	83.2208/208 peersIndependent evidence	64.9178/178 peersIndependent evidence	68.628/28 peersIndependent evidence	—	—	38.931/31 peersIndependent evidence	68.9114/114 peersIndependent evidence	86.6198/198 peersIndependent evidence	65.191/91 peersIndependent evidence	73.8202/202 peersIndependent evidence	44.9201/201 peersIndependent evidence	64.138/38 peersIndependent evidence	—	—	—	9.77/7 peersIndependent evidence	—	—	—	—	—	—	—	—	—	—	—	—	27.750/50 peersIndependent evidence	—	9.425/25 peersIndependent evidence	37.330/30 peersIndependent evidence	60.833/33 peersIndependent evidence	49.932/32 peersIndependent evidence	48.632/32 peersIndependent evidence	31.332/32 peersIndependent evidence	—	—	—	—	—	—	—	—	—	—
33	Claude Opus 4.6 · MaxAnthropic Measured effort: Max	39.2	Eligible · 100%22 published · 2 families · 1 selected capabilities	—	—	64.048/48 peersIndependent evidence	85.226/26 peersIndependent evidence	86.3197/197 peersIndependent evidence	—	89.6200/200 peersIndependent evidence	—	—	—	97.563/63 peersIndependent evidence	73.1213/213 peersIndependent evidence	—	89.5208/208 peersIndependent evidence	64.6178/178 peersIndependent evidence	—	—	—	—	77.8114/114 peersIndependent evidence	93.4198/198 peersIndependent evidence	—	87.1171/171 peersIndependent evidence	95.7169/169 peersIndependent evidence	81.538/38 peersIndependent evidence	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	52.360/60 peersIndependent evidence	52.239/39 peersIndependent evidence	45.425/25 peersIndependent evidence	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—
34	Gemini 3.1 Pro · HighGoogle Measured effort: High	38.5	Eligible · 100%45 published · 5 families · 1 selected capabilities	—	—	77.648/48 peersIndependent evidence	—	—	—	—	—	—	—	—	95.847/47 peersIndependent evidence	8.432/32 peersIndependent evidence	90.06/6 peersIndependent evidence	—	—	—	—	89.831/31 peersIndependent evidence	100.01/1 peersPublisher reports	—	—	—	8.430/30 peersIndependent evidence	—	1.625/25 peersIndependent evidence	—	—	—	47.44/4 peersIndependent evidence	—	—	—	56.526/26 peersIndependent evidence	29.526/26 peersIndependent evidence	73.226/26 peersIndependent evidence	62.026/26 peersIndependent evidence	20.526/26 peersIndependent evidence	37.526/26 peersIndependent evidence	88.726/26 peersIndependent evidence	—	85.260/60 peersIndependent evidence	—	100.06/6 peersIndependent evidence	14.430/30 peersIndependent evidence	22.733/33 peersIndependent evidence	20.132/32 peersIndependent evidence	20.132/32 peersIndependent evidence	16.832/32 peersIndependent evidence	—	—	0.02/2 peersPublisher reports	100.01/1 peersPublisher reports	—	—	—	—	—	—
35	DeepSeek V4-Pro 0424 · MaxDeepSeek Measured effort: Max	38.4	Eligible · 100%23 published · 10 families · 1 selected capabilities	58.528/28 peersIndependent evidence	67.587/87 peersIndependent evidence	—	65.726/26 peersIndependent evidence	82.6198/198 peersIndependent evidence	90.589/89 peersIndependent evidence	92.0200/200 peersIndependent evidence	61.841/41 peersIndependent evidence	75.488/88 peersIndependent evidence	81.9123/123 peersIndependent evidence	—	89.6212/212 peersIndependent evidence	70.737/37 peersIndependent evidence	92.8208/208 peersIndependent evidence	95.5178/178 peersIndependent evidence	80.728/28 peersIndependent evidence	—	—	—	—	91.5198/198 peersIndependent evidence	77.491/91 peersIndependent evidence	87.7202/202 peersIndependent evidence	86.3201/201 peersIndependent evidence	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—
36	Qwen3.7-max · ReasoningQwen Measured effort: Reasoning	35.3	Eligible · 100%21 published · 9 families · 1 selected capabilities	59.028/28 peersIndependent evidence	65.987/87 peersIndependent evidence	—	—	90.4197/197 peersIndependent evidence	67.089/89 peersIndependent evidence	91.0200/200 peersIndependent evidence	88.341/41 peersIndependent evidence	56.188/88 peersIndependent evidence	79.7123/123 peersIndependent evidence	—	94.7212/212 peersIndependent evidence	61.537/37 peersIndependent evidence	93.1208/208 peersIndependent evidence	98.4178/178 peersIndependent evidence	94.328/28 peersIndependent evidence	—	—	—	—	80.5198/198 peersIndependent evidence	70.491/91 peersIndependent evidence	71.3202/202 peersIndependent evidence	82.4201/201 peersIndependent evidence	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—
37	GPT-5.4 Mini · XHighOpenAI Measured effort: XHigh	32.7	Eligible · 100%58 published · 14 families · 1 selected capabilities	30.228/28 peersIndependent evidence	46.587/87 peersIndependent evidence	—	64.826/26 peersIndependent evidence	52.6198/198 peersIndependent evidence	64.289/89 peersIndependent evidence	86.1200/200 peersIndependent evidence	35.741/41 peersIndependent evidence	70.288/88 peersIndependent evidence	68.4123/123 peersIndependent evidence	86.263/63 peersIndependent evidence	83.0212/212 peersIndependent evidence	10.850/50 peersIndependent evidence	79.1208/208 peersIndependent evidence	88.9178/178 peersIndependent evidence	55.228/28 peersIndependent evidence	—	—	8.031/31 peersIndependent evidence	69.8114/114 peersIndependent evidence	84.4198/198 peersIndependent evidence	76.791/91 peersIndependent evidence	72.1202/202 peersIndependent evidence	73.9201/201 peersIndependent evidence	47.638/38 peersIndependent evidence	—	—	—	12.95/5 peersIndependent evidence	—	—	—	—	12.326/26 peersIndependent evidence	4.026/26 peersIndependent evidence	16.626/26 peersIndependent evidence	56.926/26 peersIndependent evidence	24.326/26 peersIndependent evidence	6.526/26 peersIndependent evidence	6.926/26 peersIndependent evidence	—	—	—	—	20.230/30 peersIndependent evidence	15.233/33 peersIndependent evidence	7.932/32 peersIndependent evidence	19.832/32 peersIndependent evidence	26.532/32 peersIndependent evidence	—	—	—	—	—	—	—	—	—	—
38	DeepSeek V4 Flash 0424 · HighDeepSeek Measured effort: High	32.5	Eligible · 100%18 published · 5 families · 1 selected capabilities	—	68.487/87 peersIndependent evidence	—	—	66.5198/198 peersIndependent evidence	68.089/89 peersIndependent evidence	80.3200/200 peersIndependent evidence	—	59.188/88 peersIndependent evidence	69.3123/123 peersIndependent evidence	—	86.1212/212 peersIndependent evidence	—	88.3208/208 peersIndependent evidence	90.6178/178 peersIndependent evidence	—	—	—	—	—	86.0198/198 peersIndependent evidence	36.291/91 peersIndependent evidence	85.3202/202 peersIndependent evidence	77.1201/201 peersIndependent evidence	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—
39	Kimi-K2.7-Code · ReasoningMoonshot Measured effort: Reasoning	31.8	Eligible · 100%20 published · 7 families · 1 selected capabilities	—	57.087/87 peersIndependent evidence	—	—	82.2198/198 peersIndependent evidence	67.889/89 peersIndependent evidence	88.4200/200 peersIndependent evidence	57.941/41 peersIndependent evidence	66.888/88 peersIndependent evidence	73.5123/123 peersIndependent evidence	—	88.9212/212 peersIndependent evidence	78.037/37 peersIndependent evidence	88.2208/208 peersIndependent evidence	74.0178/178 peersIndependent evidence	—	—	—	—	—	89.2198/198 peersIndependent evidence	65.091/91 peersIndependent evidence	76.1202/202 peersIndependent evidence	79.4201/201 peersIndependent evidence	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—
40	GLM-5.1 · ReasoningZ.ai Measured effort: Reasoning	31.7	Eligible · 100%18 published · 6 families · 1 selected capabilities	—	69.887/87 peersIndependent evidence	—	—	83.1197/197 peersIndependent evidence	62.989/89 peersIndependent evidence	82.6200/200 peersIndependent evidence	—	53.388/88 peersIndependent evidence	77.5123/123 peersIndependent evidence	—	85.9212/212 peersIndependent evidence	—	86.9208/208 peersIndependent evidence	95.2178/178 peersIndependent evidence	88.028/28 peersIndependent evidence	—	—	—	—	65.9198/198 peersIndependent evidence	54.691/91 peersIndependent evidence	76.3202/202 peersIndependent evidence	77.4201/201 peersIndependent evidence	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—
41	Hy3 · ReasoningTencent Measured effort: Reasoning	31.6	Eligible · 100%14 published · 5 families · 1 selected capabilities	—	66.487/87 peersIndependent evidence	—	—	91.7197/197 peersIndependent evidence	58.189/89 peersIndependent evidence	82.9200/200 peersIndependent evidence	—	65.788/88 peersIndependent evidence	75.6123/123 peersIndependent evidence	—	90.3212/212 peersIndependent evidence	—	87.6208/208 peersIndependent evidence	—	—	—	—	—	—	82.8198/198 peersIndependent evidence	67.691/91 peersIndependent evidence	67.6112/112 peersIndependent evidence	63.2125/125 peersIndependent evidence	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—
42	Qwen3.5-27B · ReasoningQwen Measured effort: Reasoning	31.6	Eligible · 100%12 published · 2 families · 1 selected capabilities	—	—	—	—	90.7197/197 peersIndependent evidence	—	64.9200/200 peersIndependent evidence	—	—	—	79.963/63 peersIndependent evidence	83.5212/212 peersIndependent evidence	—	79.3208/208 peersIndependent evidence	93.5178/178 peersIndependent evidence	67.028/28 peersIndependent evidence	—	—	—	81.0114/114 peersIndependent evidence	56.5198/198 peersIndependent evidence	—	89.6171/171 peersIndependent evidence	80.0169/169 peersIndependent evidence	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—
43	Qwen3.6-plus · ReasoningQwen Measured effort: Reasoning	30.4	Eligible · 100%13 published · 2 families · 1 selected capabilities	—	—	—	—	90.8197/197 peersIndependent evidence	—	78.0200/200 peersIndependent evidence	—	—	66.2123/123 peersIndependent evidence	—	87.6212/212 peersIndependent evidence	—	81.9208/208 peersIndependent evidence	92.2178/178 peersIndependent evidence	—	—	—	—	85.9114/114 peersIndependent evidence	74.1198/198 peersIndependent evidence	—	83.6202/202 peersIndependent evidence	84.6201/201 peersIndependent evidence	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—
44	Kimi-K2.6 · ReasoningMoonshot Measured effort: Reasoning	30.2	Eligible · 100%25 published · 9 families · 1 selected capabilities	—	55.787/87 peersIndependent evidence	—	83.426/26 peersIndependent evidence	96.7197/197 peersIndependent evidence	55.589/89 peersIndependent evidence	87.7200/200 peersIndependent evidence	49.641/41 peersIndependent evidence	73.788/88 peersIndependent evidence	74.4123/123 peersIndependent evidence	—	93.5212/212 peersIndependent evidence	66.037/37 peersIndependent evidence	91.8208/208 peersIndependent evidence	94.9178/178 peersIndependent evidence	45.528/28 peersIndependent evidence	—	—	—	40.6114/114 peersIndependent evidence	83.3198/198 peersIndependent evidence	79.791/91 peersIndependent evidence	84.8202/202 peersIndependent evidence	94.5201/201 peersIndependent evidence	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	50.02/2 peersPublisher reports	0.01/1 peersPublisher reports	—	—	—	—	—	—
45	Agnes 2.5 Pro Alpha · ReasoningAgnes AI Measured effort: Reasoning	30.1	Eligible · 100%14 published · 5 families · 1 selected capabilities	—	78.187/87 peersIndependent evidence	—	—	85.4197/197 peersIndependent evidence	76.089/89 peersIndependent evidence	89.8200/200 peersIndependent evidence	—	55.488/88 peersIndependent evidence	70.0123/123 peersIndependent evidence	—	86.3212/212 peersIndependent evidence	—	88.2208/208 peersIndependent evidence	—	—	—	—	—	—	84.3198/198 peersIndependent evidence	46.491/91 peersIndependent evidence	44.5112/112 peersIndependent evidence	74.6125/125 peersIndependent evidence	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—
46	MiniMax-M3 · ReasoningMiniMax Measured effort: Reasoning	30.1	Eligible · 100%22 published · 8 families · 1 selected capabilities	27.028/28 peersIndependent evidence	77.387/87 peersIndependent evidence	—	—	87.9198/198 peersIndependent evidence	65.689/89 peersIndependent evidence	79.4200/200 peersIndependent evidence	35.341/41 peersIndependent evidence	59.588/88 peersIndependent evidence	84.0123/123 peersIndependent evidence	—	96.6212/212 peersIndependent evidence	86.837/37 peersIndependent evidence	91.6208/208 peersIndependent evidence	99.9178/178 peersIndependent evidence	—	—	—	—	86.8114/114 peersIndependent evidence	37.2198/198 peersIndependent evidence	63.691/91 peersIndependent evidence	72.3202/202 peersIndependent evidence	78.9201/201 peersIndependent evidence	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—
47	Gemini 3.5 Flash-Lite · ReasoningGoogle Measured effort: Reasoning	29.7	Eligible · 100%17 published · 6 families · 1 selected capabilities	—	43.187/87 peersIndependent evidence	—	—	66.4198/198 peersIndependent evidence	68.789/89 peersIndependent evidence	25.4200/200 peersIndependent evidence	74.341/41 peersIndependent evidence	75.888/88 peersIndependent evidence	65.1123/123 peersIndependent evidence	—	74.5212/212 peersIndependent evidence	—	71.2208/208 peersIndependent evidence	—	—	—	—	—	88.9114/114 peersIndependent evidence	78.9198/198 peersIndependent evidence	40.091/91 peersIndependent evidence	57.9112/112 peersIndependent evidence	60.4125/125 peersIndependent evidence	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—
48	GPT-5.4 Nano · XHighOpenAI Measured effort: XHigh	29.3	Eligible · 100%47 published · 10 families · 1 selected capabilities	—	43.587/87 peersIndependent evidence	—	59.326/26 peersIndependent evidence	57.6198/198 peersIndependent evidence	43.989/89 peersIndependent evidence	87.0200/200 peersIndependent evidence	28.941/41 peersIndependent evidence	48.788/88 peersIndependent evidence	62.4123/123 peersIndependent evidence	—	68.8212/212 peersIndependent evidence	9.937/37 peersIndependent evidence	80.7208/208 peersIndependent evidence	93.8178/178 peersIndependent evidence	19.728/28 peersIndependent evidence	—	—	—	52.7114/114 peersIndependent evidence	70.7198/198 peersIndependent evidence	59.891/91 peersIndependent evidence	71.3202/202 peersIndependent evidence	68.9201/201 peersIndependent evidence	37.738/38 peersIndependent evidence	—	—	—	—	—	—	—	—	39.926/26 peersIndependent evidence	3.426/26 peersIndependent evidence	6.926/26 peersIndependent evidence	45.326/26 peersIndependent evidence	36.126/26 peersIndependent evidence	30.226/26 peersIndependent evidence	46.026/26 peersIndependent evidence	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—
49	Qwen3.7-plus · ReasoningQwen Measured effort: Reasoning	27.8	Eligible · 100%22 published · 8 families · 1 selected capabilities	—	64.587/87 peersIndependent evidence	—	59.926/26 peersIndependent evidence	64.6198/198 peersIndependent evidence	59.489/89 peersIndependent evidence	87.0200/200 peersIndependent evidence	66.041/41 peersIndependent evidence	70.988/88 peersIndependent evidence	54.9123/123 peersIndependent evidence	—	90.1212/212 peersIndependent evidence	51.237/37 peersIndependent evidence	88.6208/208 peersIndependent evidence	96.5178/178 peersIndependent evidence	—	—	—	—	92.8114/114 peersIndependent evidence	61.6198/198 peersIndependent evidence	60.291/91 peersIndependent evidence	77.0202/202 peersIndependent evidence	76.3201/201 peersIndependent evidence	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—
50	GLM-4.7 · ReasoningZ.ai Measured effort: Reasoning	27.5	Eligible · 100%15 published · 2 families · 1 selected capabilities	—	—	97.2108/108 peersIndependent evidence	—	77.2197/197 peersIndependent evidence	—	76.1200/200 peersIndependent evidence	—	—	72.1123/123 peersIndependent evidence	—	83.8212/212 peersIndependent evidence	—	81.3208/208 peersIndependent evidence	81.9178/178 peersIndependent evidence	—	98.5112/112 peersIndependent evidence	—	92.2114/114 peersIndependent evidence	—	78.8198/198 peersIndependent evidence	—	72.9202/202 peersIndependent evidence	67.9201/201 peersIndependent evidence	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—

About this ranking

This profile gives agentic work all the weight. Published agent-task evaluations contribute through matched benchmark families. Harnesses and reported settings remain attached to the evidence. These results describe the evaluated configurations; they do not establish how a model will perform in every workflow.

View the Capability ranking →

Behind the ranking

273 entries with published results · 127 ranked · 0 provisional estimates.

Explore seven capabilities, the original results, and gaps in the evidence.

Coding ranking Reasoning ranking Agentic ranking Compare capabilities ↗Published benchmarks ↗Evidence coverage ↗

Cite this ranking

Include the model and effort setting, the selected profile, and the dated data downloads. A shared link uses the latest data; it does not freeze a ranking in time.

UnifyBench ranking
Data updated: 12 Sept 2026
Mode: custom; evidence: documented; configurations: best
Weights: v3;agentic:100,hard_reasoning:0,coding:0,human_pref:0,knowledge:0,multimodal:0,long_context:0
https://unifybench.ai/rankings/agentic?page=2

Catalog data · Effort measurements · Configuration scoring rules · Methodology

Catalog dated 12 Sept 2026. Effort evidence collected 12 Sept 2026. Collection dates are not evaluation dates.

What does the score mean?

The adjusted comparison score estimates performance against the same reference panel from matched published comparisons. Capability pools matched comparisons across families and checks aggregate breadth and independent corroboration. Custom profiles require support in every selected capability. Missing results never count as losses.

Family details show observed win shares against available reference peers. Those descriptive values are supporting evidence, not adjusted capability scores. Provider reports and independent results remain labeled. Matching reported settings do not establish equal inference effort, and score differences do not establish statistical significance.

Read the full methodology →