Reasoning benchmark / ARC Prize

ARC-AGI-2

Lumina Verified

Verified abstract reasoning results on ARC-AGI-2.

Version 2Verified 31 October 2026279 published resultsOfficial source

ARC-AGI-2 · Lumina Score

Semi-Private · verified model evaluations

LuminaBench

20 of 279 configurations

Lumina Score vs Cost per Task

Publisher-reported USD per task.

LuminaBench

15 of 95 families

Missing resource measurements remain unavailable. Missing metrics are omitted from their graph; quality scores remain available.

Results table

ARC-AGI-2 2. Original publisher values and provisional Lumina conversion.
Rank
1	GPT-6 AstraMax	1,135	95%	$1.12
2	GPT-6.1 SolMax	1,133	94.17%	$0.25
3	Claude Opus 5.5High	1,131	93.33%	$0.41
3	GPT-6 AstraxHigh	1,131	93.33%	$0.83
5	Claude Opus 5.5xHigh	1,129	92.5%	$0.67
5	GPT-5.6 SolMax	1,129	92.5%	$1.44
7	GPT-6 AstraHigh	1,128	92.08%	$0.67
7	GPT-6 AstraMedium	1,128	92.08%	$0.48
7	Claude Opus 5.5Max	1,128	91.67%	$1.85
7	GPT-6.1 SolHigh	1,128	91.67%	$0.13
7	GPT-6.1 SolxHigh	1,128	91.67%	$0.18
12	Gemini 4 ArgonHigh	1,127	91.25%	$0.32
13	Claude Opus 5Max	1,125	90.42%	$2.06
14	Claude Fable 5.1Max	1,124	90%	$4.49
14	Claude Fable 5.1xHigh	1,124	90%	$3.12
14	GPT-5.6 SolxHigh	1,124	90%	$1.04
17	GPT-6 SolMax	1,123	89.58%	$0.44
18	Claude Fable 5Max	1,122	89.17%	$5.45
18	Gemini 3.8 FlashHigh	1,122	89.17%	$0.4
20	Claude Fable 5.1High	1,121	88.75%	$1.67
21	Claude Fable 5xHigh	1,120	88.33%	$3
21	Claude Opus 5High	1,120	88.33%	$1.45
23	Claude Fable 5High	1,118	87.5%	$2.44
23	Claude Opus 5.5Medium	1,118	87.5%	$0.34
25	Gemini 4 ArgonMedium	1,117	87.08%	$0.14
26	GPT-6.1 SolMedium	1,116	86.67%	$0.1
27	Claude Fable 5.1Medium	1,115	86.25%	$1.22
28	GPT-5.6 SolHigh	1,114	85.42%	$0.74
29	GPT-6 AstraLow	1,113	85.42%	$0.42
29	GPT-5.5xHigh	1,113	85%	$1.87
31	Gemini 3.7 FlashHigh	1,112	84.58%	$0.25
31	Gemini 3 Deep Think (2/26)Not specified	1,112	84.58%	$13.62
31	GPT-5.5 ProHigh	1,112	84.58%	$10.51
34	GPT-5.5 ProxHigh	1,111	84.16%	$10.76
35	GPT-5.6 TerraMax	1,110	83.9%	$1.09
35	Claude Sonnet 5.5xHigh	1,110	83.89%	$0.55
37	GPT-5.4 ProxHigh	1,109	83.33%	$16.41
37	GPT-5.5High	1,109	83.33%	$1.45
39	Gemini 3.8 FlashMedium	1,108	82.92%	$0.25
40	Claude Fable 5Medium	1,107	82.5%	$1.93
41	Claude Fable 5.1Low	1,098	78.33%	$0.95
42	DeepSeek V4.1 Flash 1017Max	1,097	78.06%	$0.16
42	GPT-6 SolxHigh	1,097	78.06%	$0.28
44	Gemini 3.8 FlashLow	1,096	77.5%	$0.15
45	Gemini 3.1 Pro (Preview)Not specified	1,095	77.08%	$0.96
46	Dots3-Note PreviewMax	1,094	76.81%	$0.08
46	Claude Fable 5Low	1,094	76.81%	$1.45
46	GPT-6.1 SolLow	1,094	76.67%	$0.09
49	Claude 4.7Max	1,092	75.83%	$7.43
50	GPT-5.6 TerraxHigh	1,088	74.17%	$0.69
50	GPT-5.4xHigh	1,088	73.95%	$1.52
52	DeepSeek V4.1 Flash 1017Low	1,087	73.61%	$0.13
53	DeepSeek V4.1 FlashMax	1,085	72.92%	$0.13
54	Claude Opus 4.8High	1,084	72.08%	$2.74
54	Gemini 3.5 FlashHigh	1,084	72.08%	$0.85
56	Claude Opus 4.8Medium	1,083	71.67%	$2.39
57	GPT-5.5Medium	1,080	70.42%	$0.86
58	Claude Opus 5.5Low	1,079	70.14%	$0.24
59	Claude Opus 4.6120K, High	1,077	69.17%	$3.47
60	GPT-6 SolHigh	1,076	68.89%	$0.2
60	Claude Opus 4.6120K, Max	1,076	68.75%	$3.64
60	DeepSeek V4.1 Flash 1017High	1,076	68.61%	$0.15
63	Claude 4.7High	1,075	68.33%	$3.17
64	Claude 4.7Medium	1,073	67.5%	$2.96
64	DeepSeek V4.1 FlashHigh	1,073	67.5%	$0.13
64	GPT-5.4High	1,073	67.5%	$1.02
64	Claude Sonnet 5.5High	1,073	67.36%	$0.28
68	Grok 4.6xHigh	1,072	67.08%	$0.76
68	GPT-5.6 SolMedium	1,072	67.08%	$0.47
68	GPT-5.6 TerraHigh	1,072	67.08%	$0.55
71	Claude Opus 4.6120K, Medium	1,070	66.25%	$2.72
72	GLM 5.3 FlashMax	1,069	65.83%	$0.09
73	Grok 4.20 (Reasoning)Not specified	1,068	65.14%	$0.92
73	Grok 4.6High	1,068	65.14%	$0.73
75	Claude Opus 4.6120K, Low	1,067	64.58%	$2.25
76	Gemini 3.7 FlashMedium	1,065	63.75%	$0.12
77	Claude Opus 4.8Low	1,061	62.22%	$1.68
77	Claude 4.7Low	1,061	62.08%	$2.38
79	DeepSeek V4 Flash 0731Max	1,059	61.39%	$0.04
79	Grok 4.7xHigh	1,059	61.39%	$2.01
79	Claude Haiku 5.5xHigh	1,059	61.25%	$0.08
79	DeepSeek V4 Pro 0813Max	1,059	61.25%	$0.6
79	Grok 4.6Medium	1,059	61.25%	$0.66
84	DeepSeek V4.1 FlashLow	1,058	60.56%	$0.12
85	Claude Sonnet 4.6High	1,057	60.42%	$2.7
85	Gemini 3.6 FlashHigh	1,057	60.42%	$0.61
85	Kimi K3Max	1,057	60.42%	$1.59
88	DeepSeek V4 Pro 0813High	1,056	59.72%	$0.6
89	GPT-5.6 Luna 2026-07-30Max	1,055	59.58%	$0.18
89	GPT-6 AstraNone	1,055	59.58%	$0.37
89	GPT-5.6 LunaMax	1,055	59.54%	$0.67
89	GPT-6 LunaMax	1,055	59.31%	$0.06
93	Grok 4.7Medium	1,054	58.75%	$1.48
94	Grok 4.7High	1,053	58.33%	$1.55
94	Claude Sonnet 4.6Max	1,053	58.33%	$2.72
96	GPT-6 SolMedium	1,051	57.78%	$0.15
96	Muse Spark OpenxHigh	1,051	57.5%	$0.73
98	DeepSeek V4 Pro 0813Low	1,048	56.25%	$0.37
99	DeepSeek V4 Flash 0731High	1,047	55.97%	$0.05
100	GPT-5.4Medium	1,046	55.42%	$0.68
101	Kimi K3High	1,045	55%	$0.95
102	GPT-5.2 ProHigh	1,043	54.16%	$15.72
103	Gemini 3.7 FlashLow	1,040	52.92%	$0.08
103	GPT-5.2xHigh	1,040	52.91%	$1.9
103	Grok 4.5High	1,040	52.64%	$0.78
103	Grok 4.5Medium	1,040	52.64%	$0.75
107	Gemini 3.6 FlashMedium	1,035	50.42%	$0.46
108	GLM 5.3 FlashHigh	1,034	50.14%	$0.08
109	GPT-5.6 LunaxHigh	1,029	47.64%	$0.54
110	Muse Spark OpenHigh	1,028	47.5%	$0.62
111	DeepSeek V4 Flash 0731Low	1,025	45.97%	$0.02
112	GPT-5.6 Luna 2026-07-30xHigh	1,019	43.47%	$0.13
112	GPT-5.2High	1,019	43.33%	$1.39
114	GPT-5.6 SolLow	1,017	42.5%	$0.32
114	Qwen3.8-27BxHigh	1,017	42.36%	$0.45
116	Gemini 4 ArgonLow	1,016	42.08%	$0.07
116	GPT-6 LunaxHigh	1,016	41.94%	$0.03
118	Inkling SmallxHigh	1,012	40.14%	$0.23
119	GPT-5.2 ProMedium	1,008	38.47%	$8.99
120	Claude Opus 4.5Thinking, 64K	1,006	37.64%	$2.4
120	GPT-5.6 TerraMedium	1,006	37.5%	$0.27
122	InklingNot specified	1,004	36.53%	$0.64
123	Gemini 3 Flash PreviewHigh	997	33.61%	$0.23
124	GPT-5.6 Luna 2026-07-30High	996	33.33%	$0.06
124	GPT-5.5Low	996	33.33%	$0.35
124	Grok 4.5Low	996	33.06%	$0.4
124	Inkling SmallHigh	996	33.06%	$0.13
128	Claude Sonnet 5.5Medium	995	32.78%	$0.16
129	GPT-6 SolLow	992	31.53%	$0.1
129	GPT-6 LunaHigh	992	31.39%	$0.02
131	Gemini 3 ProNot specified	991	31.11%	$0.81
132	Claude Haiku 5.5High	990	30.69%	$0.05
132	Gemini 3.6 FlashLow	990	30.42%	$0.22
134	GPT-5.6 LunaHigh	987	29.31%	$0.29
134	GPT-5.4Low	987	29.17%	$0.27
136	GLM 5.3 FlashLow	984	27.92%	$0.06
136	Grok 4.6Low	984	27.64%	$0.21
138	Muse Spark OpenMedium	983	27.5%	$0.25
139	GPT-5.2Medium	981	26.67%	$0.76
140	Claude Sonnet 5.5Low	975	24.03%	$0.11
141	Beam 1.0xHigh	974	23.47%	$0.61
142	Claude Opus 4.5Thinking, 16K	973	22.78%	$0.79
142	Qwen3.8-27BLow	973	22.78%	$0.24
142	GLM-5.2Not specified	973	22.78%	$0.25
145	Beam 1.0High	966	20%	$0.35
146	GPT-5.4 MinixHigh	964	18.9%	$0.75
146	GPT-5.6 TerraLow	964	18.75%	$0.17
148	GPT-5 ProNot specified	963	18.33%	$7.14
149	GPT-6 LunaMedium	962	18.06%	$0.01
150	GPT-5.1Thinking, High	961	17.64%	$1.17
151	Grok 4.7Low	959	16.67%	$0.17
152	Grok 4Thinking	957	15.98%	$2.17
153	Claude Opus 4.5Thinking, 8K	953	13.89%	$0.48
154	Inkling SmallMedium	952	13.61%	$0.06
154	Claude Sonnet 4.5Thinking 32K	952	13.61%	$0.76
156	Qwen3.8-27BMedium	951	13.19%	$0.29
156	GPT-5.4 MiniHigh	951	13.19%	$0.56
158	Gemini 3 Flash PreviewMedium	950	12.78%	$0.08
159	Kimi K3Low	949	12.36%	$0.25
159	Muse Spark OpenLow	949	12.22%	$0.11
161	Kimi K2.5Not specified	948	11.81%	$0.28
162	Gemini 3.5 Flash-LiteHigh	945	10.28%	$0.14
163	GPT-5High	944	9.86%	$0.73
164	GPT-5.2Low	943	9.72%	$0.26
164	Claude Haiku 5.5Medium	943	9.44%	$0.01
166	Gemini 3.5 FlashMinimal	941	8.89%	$0.11
166	Claude Opus 4Thinking 16K	941	8.61%	$1.93
168	Claude Opus 4.5Thinking, None	939	7.78%	$0.22
169	GPT-5Medium	938	7.49%	$0.45
169	GPT-5.6 Luna 2026-07-30Medium	938	7.36%	$0.02
169	GPT-5.6 LunaMedium	938	7.36%	$0.11
172	Claude Sonnet 4.5Thinking 16K	937	6.94%	$0.35
172	Claude Sonnet 4.5Thinking 8K	937	6.94%	$0.23
174	GPT-5.1Thinking, Medium	936	6.53%	$0.42
174	o3High	936	6.53%	$0.83
174	Beam 1.0Medium	936	6.39%	$0.18
177	o4-miniHigh	935	6.11%	$0.86
177	Claude Sonnet 4Thinking 16K	935	5.93%	$0.49
177	Claude Sonnet 4.5Thinking 1K	935	5.83%	$0.14
180	GPT-5.4 NanoxHigh	934	5.69%	$0.16
181	Grok 4 (Fast Reasoning)Not specified	933	5.28%	$0.06
181	Gemini 3.5 Flash-LiteMedium	933	5.28%	$0.04
181	GPT-5.6 LunaLow	933	5.14%	$0.07
184	Inkling SmallLow	932	4.86%	$0.06
184	Gemini 2.5 ProThinking 32K	932	4.86%	$0.76
184	GLM-5Not specified	932	4.86%	$0.27
184	Minimax M2.5Not specified	932	4.86%	$0.17
184	GPT-6 LunaLow	932	4.58%	$0.01
184	Claude Opus 4Thinking 8K	932	4.52%	$1.16
190	GPT-5.4 MiniMedium	931	4.44%	$0.33
190	GPT-5 MiniHigh	931	4.44%	$0.2
190	Claude Haiku 4.5Thinking 32K	931	4.03%	$0.38
190	Deepseek V3.2Not specified	931	4.03%	$0.12
190	Gemini 2.5 ProThinking 16K	931	4.03%	$0.71
190	GPT-5 MiniMedium	931	4.03%	$0.06
190	Inkling SmallMinimal	931	4.03%	$0.05
197	Claude Sonnet 4.5Not specified	930	3.75%	$0.13
197	GPT-5.6 Luna 2026-07-30Low	930	3.61%	$0.01
197	GPT-5.4 NanoHigh	930	3.61%	$0.13
200	Gemini 3 Flash PreviewMinimal	929	3.33%	$0.02
201	o3-miniHigh	928	2.99%	$0.55
201	o3Medium	928	2.98%	$0.48
201	Gemini 2.5 ProThinking 8K	928	2.92%	$0.44
201	Claude Haiku 4.5Thinking 16K	928	2.78%	$0.14
201	Beam 1.0Low	928	2.78%	$0.1
201	Claude Haiku 5.5Low	928	2.78%	$0.01
207	Gemini 3.6 FlashMinimal	927	2.64%	$0.07
207	GPT-5 NanoHigh	927	2.61%	$0.03
207	Gemini 2.5 Flash (Preview)Thinking 24K	927	2.54%	$0.32
207	o4-miniMedium	927	2.36%	$0.23
211	Gemini 2.5 Flash (Preview)Thinking 1K	926	2.16%	$0.03
211	Claude Sonnet 4Thinking 8K	926	2.12%	$0.27
211	Gemini 2.5 Flash (Preview)Thinking 8K	926	2.12%	$0.2
211	DeepSeek V4.1 Flash 1017None	926	2.08%	$0.01
211	DeepSeek V4 Flash 0731None	926	2.08%	$0
211	o3-miniMedium	926	2.08%	$0.28
211	o3Low	926	1.99%	$0.23
211	Gemini 2.5 Flash (Preview)Thinking 16K	926	1.98%	$0.32
211	Nemotron 3.5 Super VLMedium	926	1.94%	$0.02
211	GPT-5.1Thinking, Low	926	1.94%	$0.13
211	GPT-5Low	926	1.94%	$0.19
211	GPT-5.4 NanoMedium	926	1.94%	$0.06
223	Gemini 2.5 Flash (Preview)Not specified	925	1.69%	$0.06
223	Claude Haiku 4.5Thinking 8K	925	1.67%	$0.09
223	GPT-5 MiniMinimal	925	1.67%	$0.01
223	o4-miniLow	925	1.67%	$0.05
223	GPT-6 SolNone	925	1.67%	$0.07
223	GPT-5.4 NanoLow	925	1.53%	$0.01
223	Qwen3.8-27BNone	925	1.53%	$0.06
223	Gemini 3.5 Flash-LiteLow	925	1.53%	$0.02
223	Nemotron 3.5 Super VLHigh	925	1.53%	$0.04
232	Gemini 2.0 FlashNot specified	924	1.3%	$0
232	Deepseek R1Not specified	924	1.3%	$0.08
232	Claude Opus 4Not specified	924	1.27%	$0.64
232	Claude Sonnet 4Not specified	924	1.27%	$0.13
232	Codex Mini (Latest)Not specified	924	1.27%	$0.23
232	Claude Haiku 4.5Not specified	924	1.25%	$0.04
232	Claude Haiku 4.5Thinking 1K	924	1.25%	$0.05
232	DeepSeek V4.1 FlashNone	924	1.25%	$0.01
232	Gemini 3 Flash PreviewLow	924	1.25%	$0.02
232	Qwen3-235b-a22b Instruct (25/07)Not specified	924	1.25%	$0
232	Deepseek R1 (05/28)Not specified	924	1.12%	$0.05
232	GPT-5.4 MiniLow	924	1.11%	$0.06
244	Claude 3.78K	923	0.9%	$0.36
244	GPT-5 NanoMedium	923	0.88%	$0.01
244	Claude Sonnet 4Thinking 1K	923	0.85%	$0.14
244	DeepSeek V4 Pro 0813None	923	0.83%	$0.04
244	Gemini 3.5 Flash-LiteMinimal	923	0.83%	$0.01
244	Nemotron 3.5 Super VLNone	923	0.83%	$0.01
244	GPT-5.6 Luna 2026-07-30None	923	0.83%	$0.01
244	GPT-5.2Not specified	923	0.83%	$0.08
244	GPT-5 MiniLow	923	0.83%	$0.02
244	o1-miniNot specified	923	0.83%	$0.19
244	Gemini 1.5 ProNot specified	923	0.8%	$0.04
244	GPT-4.5Not specified	923	0.8%	$2.1
244	Claude 3.716K	923	0.7%	$0.51
257	GPT-4.1Not specified	922	0.42%	$0.07
257	GPT-5.1Thinking, None	922	0.42%	$0.06
257	Grok 3 MiniLow	922	0.42%	$0.01
257	Claude 3.71K	922	0.4%	$0.14
261	Claude 3.7Not specified	921	0%	$0.12
261	Claude Opus 4Thinking 1K	921	0%	$0.75
261	Gemini 2.5 ProThinking 1K	921	0%	$0.09
261	GPT-4.1-MiniNot specified	921	0%	$0.01
261	GPT-4.1-NanoNot specified	921	0%	$0
261	GPT-4oNot specified	921	0%	$0.08
261	GPT-4o-miniNot specified	921	0%	$0.01
261	GPT-5Minimal	921	0%	$0.06
261	GPT-5 NanoLow	921	0%	$0
261	GPT-5 NanoMinimal	921	0%	$0
261	Grok 3Not specified	921	0%	$0.14
261	Llama 4 MaverickNot specified	921	0%	$0.01
261	Llama 4 ScoutNot specified	921	0%	$0.01
261	Magistral MediumNot specified	921	0%	$0.11
261	Magistral MediumThinking	921	0%	$0.12
261	Magistral SmallNot specified	921	0%	$0.05
261	o3-miniLow	921	0%	$0.06
261	GPT-6 LunaNone	921	0%	$0
261	Inkling SmallNone	921	0%	$0.01

1 · GPT-6 AstraMax1135

Accuracy: 95%
USD/task: $1.12
Evidence: Model profile ↗

2 · GPT-6.1 SolMax1133

Accuracy: 94.17%
USD/task: $0.25
Evidence: Model profile ↗

3 · Claude Opus 5.5High1131

Accuracy: 93.33%
USD/task: $0.41
Evidence: Model profile ↗

3 · GPT-6 AstraxHigh1131

Accuracy: 93.33%
USD/task: $0.83
Evidence: Model profile ↗

5 · Claude Opus 5.5xHigh1129

Accuracy: 92.5%
USD/task: $0.67
Evidence: Model profile ↗

5 · GPT-5.6 SolMax1129

Accuracy: 92.5%
USD/task: $1.44
Evidence: Model profile ↗

7 · GPT-6 AstraHigh1128

Accuracy: 92.08%
USD/task: $0.67
Evidence: Model profile ↗

7 · GPT-6 AstraMedium1128

Accuracy: 92.08%
USD/task: $0.48
Evidence: Model profile ↗

7 · Claude Opus 5.5Max1128

Accuracy: 91.67%
USD/task: $1.85
Evidence: Model profile ↗

7 · GPT-6.1 SolHigh1128

Accuracy: 91.67%
USD/task: $0.13
Evidence: Model profile ↗

7 · GPT-6.1 SolxHigh1128

Accuracy: 91.67%
USD/task: $0.18
Evidence: Model profile ↗

12 · Gemini 4 ArgonHigh1127

Accuracy: 91.25%
USD/task: $0.32
Evidence: Model profile ↗

13 · Claude Opus 5Max1125

Accuracy: 90.42%
USD/task: $2.06
Evidence: Model profile ↗

14 · Claude Fable 5.1Max1124

Accuracy: 90%
USD/task: $4.49
Evidence: Model profile ↗

14 · Claude Fable 5.1xHigh1124

Accuracy: 90%
USD/task: $3.12
Evidence: Model profile ↗

14 · GPT-5.6 SolxHigh1124

Accuracy: 90%
USD/task: $1.04
Evidence: Model profile ↗

17 · GPT-6 SolMax1123

Accuracy: 89.58%
USD/task: $0.44
Evidence: Model profile ↗

18 · Claude Fable 5Max1122

Accuracy: 89.17%
USD/task: $5.45
Evidence: Model profile ↗

18 · Gemini 3.8 FlashHigh1122

Accuracy: 89.17%
USD/task: $0.4
Evidence: Model profile ↗

20 · Claude Fable 5.1High1121

Accuracy: 88.75%
USD/task: $1.67
Evidence: Model profile ↗

21 · Claude Fable 5xHigh1120

Accuracy: 88.33%
USD/task: $3
Evidence: Model profile ↗

21 · Claude Opus 5High1120

Accuracy: 88.33%
USD/task: $1.45
Evidence: Model profile ↗

23 · Claude Fable 5High1118

Accuracy: 87.5%
USD/task: $2.44
Evidence: Model profile ↗

23 · Claude Opus 5.5Medium1118

Accuracy: 87.5%
USD/task: $0.34
Evidence: Model profile ↗

25 · Gemini 4 ArgonMedium1117

Accuracy: 87.08%
USD/task: $0.14
Evidence: Model profile ↗

26 · GPT-6.1 SolMedium1116

Accuracy: 86.67%
USD/task: $0.1
Evidence: Model profile ↗

27 · Claude Fable 5.1Medium1115

Accuracy: 86.25%
USD/task: $1.22
Evidence: Model profile ↗

28 · GPT-5.6 SolHigh1114

Accuracy: 85.42%
USD/task: $0.74
Evidence: Model profile ↗

29 · GPT-6 AstraLow1113

Accuracy: 85.42%
USD/task: $0.42
Evidence: Model profile ↗

29 · GPT-5.5xHigh1113

Accuracy: 85%
USD/task: $1.87
Evidence: Model profile ↗

31 · Gemini 3.7 FlashHigh1112

Accuracy: 84.58%
USD/task: $0.25
Evidence: Model profile ↗

31 · Gemini 3 Deep Think (2/26)Not specified1112

Accuracy: 84.58%
USD/task: $13.62
Evidence: Model profile ↗

31 · GPT-5.5 ProHigh1112

Accuracy: 84.58%
USD/task: $10.51
Evidence: Model profile ↗

34 · GPT-5.5 ProxHigh1111

Accuracy: 84.16%
USD/task: $10.76
Evidence: Model profile ↗

35 · GPT-5.6 TerraMax1110

Accuracy: 83.9%
USD/task: $1.09
Evidence: Model profile ↗

35 · Claude Sonnet 5.5xHigh1110

Accuracy: 83.89%
USD/task: $0.55
Evidence: Model profile ↗

37 · GPT-5.4 ProxHigh1109

Accuracy: 83.33%
USD/task: $16.41
Evidence: Model profile ↗

37 · GPT-5.5High1109

Accuracy: 83.33%
USD/task: $1.45
Evidence: Model profile ↗

39 · Gemini 3.8 FlashMedium1108

Accuracy: 82.92%
USD/task: $0.25
Evidence: Model profile ↗

40 · Claude Fable 5Medium1107

Accuracy: 82.5%
USD/task: $1.93
Evidence: Model profile ↗

41 · Claude Fable 5.1Low1098

Accuracy: 78.33%
USD/task: $0.95
Evidence: Model profile ↗

42 · DeepSeek V4.1 Flash 1017Max1097

Accuracy: 78.06%
USD/task: $0.16
Evidence: Model profile ↗

42 · GPT-6 SolxHigh1097

Accuracy: 78.06%
USD/task: $0.28
Evidence: Model profile ↗

44 · Gemini 3.8 FlashLow1096

Accuracy: 77.5%
USD/task: $0.15
Evidence: Model profile ↗

45 · Gemini 3.1 Pro (Preview)Not specified1095

Accuracy: 77.08%
USD/task: $0.96
Evidence: Model profile ↗

46 · Dots3-Note PreviewMax1094

Accuracy: 76.81%
USD/task: $0.08
Evidence: Model profile ↗

46 · Claude Fable 5Low1094

Accuracy: 76.81%
USD/task: $1.45
Evidence: Model profile ↗

46 · GPT-6.1 SolLow1094

Accuracy: 76.67%
USD/task: $0.09
Evidence: Model profile ↗

49 · Claude 4.7Max1092

Accuracy: 75.83%
USD/task: $7.43
Evidence: Model profile ↗

50 · GPT-5.6 TerraxHigh1088

Accuracy: 74.17%
USD/task: $0.69
Evidence: Model profile ↗

50 · GPT-5.4xHigh1088

Accuracy: 73.95%
USD/task: $1.52
Evidence: Model profile ↗

52 · DeepSeek V4.1 Flash 1017Low1087

Accuracy: 73.61%
USD/task: $0.13
Evidence: Model profile ↗

53 · DeepSeek V4.1 FlashMax1085

Accuracy: 72.92%
USD/task: $0.13
Evidence: Model profile ↗

54 · Claude Opus 4.8High1084

Accuracy: 72.08%
USD/task: $2.74
Evidence: Model profile ↗

54 · Gemini 3.5 FlashHigh1084

Accuracy: 72.08%
USD/task: $0.85
Evidence: Model profile ↗

56 · Claude Opus 4.8Medium1083

Accuracy: 71.67%
USD/task: $2.39
Evidence: Model profile ↗

57 · GPT-5.5Medium1080

Accuracy: 70.42%
USD/task: $0.86
Evidence: Model profile ↗

58 · Claude Opus 5.5Low1079

Accuracy: 70.14%
USD/task: $0.24
Evidence: Model profile ↗

59 · Claude Opus 4.6120K, High1077

Accuracy: 69.17%
USD/task: $3.47
Evidence: Model profile ↗

60 · GPT-6 SolHigh1076

Accuracy: 68.89%
USD/task: $0.2
Evidence: Model profile ↗

60 · Claude Opus 4.6120K, Max1076

Accuracy: 68.75%
USD/task: $3.64
Evidence: Model profile ↗

60 · DeepSeek V4.1 Flash 1017High1076

Accuracy: 68.61%
USD/task: $0.15
Evidence: Model profile ↗

63 · Claude 4.7High1075

Accuracy: 68.33%
USD/task: $3.17
Evidence: Model profile ↗

64 · Claude 4.7Medium1073

Accuracy: 67.5%
USD/task: $2.96
Evidence: Model profile ↗

64 · DeepSeek V4.1 FlashHigh1073

Accuracy: 67.5%
USD/task: $0.13
Evidence: Model profile ↗

64 · GPT-5.4High1073

Accuracy: 67.5%
USD/task: $1.02
Evidence: Model profile ↗

64 · Claude Sonnet 5.5High1073

Accuracy: 67.36%
USD/task: $0.28
Evidence: Model profile ↗

68 · Grok 4.6xHigh1072

Accuracy: 67.08%
USD/task: $0.76
Evidence: Model profile ↗

68 · GPT-5.6 SolMedium1072

Accuracy: 67.08%
USD/task: $0.47
Evidence: Model profile ↗

68 · GPT-5.6 TerraHigh1072

Accuracy: 67.08%
USD/task: $0.55
Evidence: Model profile ↗

71 · Claude Opus 4.6120K, Medium1070

Accuracy: 66.25%
USD/task: $2.72
Evidence: Model profile ↗

72 · GLM 5.3 FlashMax1069

Accuracy: 65.83%
USD/task: $0.09
Evidence: Model profile ↗

73 · Grok 4.20 (Reasoning)Not specified1068

Accuracy: 65.14%
USD/task: $0.92
Evidence: Model profile ↗

73 · Grok 4.6High1068

Accuracy: 65.14%
USD/task: $0.73
Evidence: Model profile ↗

75 · Claude Opus 4.6120K, Low1067

Accuracy: 64.58%
USD/task: $2.25
Evidence: Model profile ↗

76 · Gemini 3.7 FlashMedium1065

Accuracy: 63.75%
USD/task: $0.12
Evidence: Model profile ↗

77 · Claude Opus 4.8Low1061

Accuracy: 62.22%
USD/task: $1.68
Evidence: Model profile ↗

77 · Claude 4.7Low1061

Accuracy: 62.08%
USD/task: $2.38
Evidence: Model profile ↗

79 · DeepSeek V4 Flash 0731Max1059

Accuracy: 61.39%
USD/task: $0.04
Evidence: Model profile ↗

79 · Grok 4.7xHigh1059

Accuracy: 61.39%
USD/task: $2.01
Evidence: Model profile ↗

79 · Claude Haiku 5.5xHigh1059

Accuracy: 61.25%
USD/task: $0.08
Evidence: Model profile ↗

79 · DeepSeek V4 Pro 0813Max1059

Accuracy: 61.25%
USD/task: $0.6
Evidence: Model profile ↗

79 · Grok 4.6Medium1059

Accuracy: 61.25%
USD/task: $0.66
Evidence: Model profile ↗

84 · DeepSeek V4.1 FlashLow1058

Accuracy: 60.56%
USD/task: $0.12
Evidence: Model profile ↗

85 · Claude Sonnet 4.6High1057

Accuracy: 60.42%
USD/task: $2.7
Evidence: Model profile ↗

85 · Gemini 3.6 FlashHigh1057

Accuracy: 60.42%
USD/task: $0.61
Evidence: Model profile ↗

85 · Kimi K3Max1057

Accuracy: 60.42%
USD/task: $1.59
Evidence: Model profile ↗

88 · DeepSeek V4 Pro 0813High1056

Accuracy: 59.72%
USD/task: $0.6
Evidence: Model profile ↗

89 · GPT-5.6 Luna 2026-07-30Max1055

Accuracy: 59.58%
USD/task: $0.18
Evidence: Model profile ↗

89 · GPT-6 AstraNone1055

Accuracy: 59.58%
USD/task: $0.37
Evidence: Model profile ↗

89 · GPT-5.6 LunaMax1055

Accuracy: 59.54%
USD/task: $0.67
Evidence: Model profile ↗

89 · GPT-6 LunaMax1055

Accuracy: 59.31%
USD/task: $0.06
Evidence: Model profile ↗

93 · Grok 4.7Medium1054

Accuracy: 58.75%
USD/task: $1.48
Evidence: Model profile ↗

94 · Grok 4.7High1053

Accuracy: 58.33%
USD/task: $1.55
Evidence: Model profile ↗

94 · Claude Sonnet 4.6Max1053

Accuracy: 58.33%
USD/task: $2.72
Evidence: Model profile ↗

96 · GPT-6 SolMedium1051

Accuracy: 57.78%
USD/task: $0.15
Evidence: Model profile ↗

96 · Muse Spark OpenxHigh1051

Accuracy: 57.5%
USD/task: $0.73
Evidence: Model profile ↗

98 · DeepSeek V4 Pro 0813Low1048

Accuracy: 56.25%
USD/task: $0.37
Evidence: Model profile ↗

99 · DeepSeek V4 Flash 0731High1047

Accuracy: 55.97%
USD/task: $0.05
Evidence: Model profile ↗

100 · GPT-5.4Medium1046

Accuracy: 55.42%
USD/task: $0.68
Evidence: Model profile ↗

101 · Kimi K3High1045

Accuracy: 55%
USD/task: $0.95
Evidence: Model profile ↗

102 · GPT-5.2 ProHigh1043

Accuracy: 54.16%
USD/task: $15.72
Evidence: Model profile ↗

103 · Gemini 3.7 FlashLow1040

Accuracy: 52.92%
USD/task: $0.08
Evidence: Model profile ↗

103 · GPT-5.2xHigh1040

Accuracy: 52.91%
USD/task: $1.9
Evidence: Model profile ↗

103 · Grok 4.5High1040

Accuracy: 52.64%
USD/task: $0.78
Evidence: Model profile ↗

103 · Grok 4.5Medium1040

Accuracy: 52.64%
USD/task: $0.75
Evidence: Model profile ↗

107 · Gemini 3.6 FlashMedium1035

Accuracy: 50.42%
USD/task: $0.46
Evidence: Model profile ↗

108 · GLM 5.3 FlashHigh1034

Accuracy: 50.14%
USD/task: $0.08
Evidence: Model profile ↗

109 · GPT-5.6 LunaxHigh1029

Accuracy: 47.64%
USD/task: $0.54
Evidence: Model profile ↗

110 · Muse Spark OpenHigh1028

Accuracy: 47.5%
USD/task: $0.62
Evidence: Model profile ↗

111 · DeepSeek V4 Flash 0731Low1025

Accuracy: 45.97%
USD/task: $0.02
Evidence: Model profile ↗

112 · GPT-5.6 Luna 2026-07-30xHigh1019

Accuracy: 43.47%
USD/task: $0.13
Evidence: Model profile ↗

112 · GPT-5.2High1019

Accuracy: 43.33%
USD/task: $1.39
Evidence: Model profile ↗

114 · GPT-5.6 SolLow1017

Accuracy: 42.5%
USD/task: $0.32
Evidence: Model profile ↗

114 · Qwen3.8-27BxHigh1017

Accuracy: 42.36%
USD/task: $0.45
Evidence: Model profile ↗

116 · Gemini 4 ArgonLow1016

Accuracy: 42.08%
USD/task: $0.07
Evidence: Model profile ↗

116 · GPT-6 LunaxHigh1016

Accuracy: 41.94%
USD/task: $0.03
Evidence: Model profile ↗

118 · Inkling SmallxHigh1012

Accuracy: 40.14%
USD/task: $0.23
Evidence: Model profile ↗

119 · GPT-5.2 ProMedium1008

Accuracy: 38.47%
USD/task: $8.99
Evidence: Model profile ↗

120 · Claude Opus 4.5Thinking, 64K1006

Accuracy: 37.64%
USD/task: $2.4
Evidence: Model profile ↗

120 · GPT-5.6 TerraMedium1006

Accuracy: 37.5%
USD/task: $0.27
Evidence: Model profile ↗

122 · InklingNot specified1004

Accuracy: 36.53%
USD/task: $0.64
Evidence: Model profile ↗

123 · Gemini 3 Flash PreviewHigh997

Accuracy: 33.61%
USD/task: $0.23
Evidence: Model profile ↗

124 · GPT-5.6 Luna 2026-07-30High996

Accuracy: 33.33%
USD/task: $0.06
Evidence: Model profile ↗

124 · GPT-5.5Low996

Accuracy: 33.33%
USD/task: $0.35
Evidence: Model profile ↗

124 · Grok 4.5Low996

Accuracy: 33.06%
USD/task: $0.4
Evidence: Model profile ↗

124 · Inkling SmallHigh996

Accuracy: 33.06%
USD/task: $0.13
Evidence: Model profile ↗

128 · Claude Sonnet 5.5Medium995

Accuracy: 32.78%
USD/task: $0.16
Evidence: Model profile ↗

129 · GPT-6 SolLow992

Accuracy: 31.53%
USD/task: $0.1
Evidence: Model profile ↗

129 · GPT-6 LunaHigh992

Accuracy: 31.39%
USD/task: $0.02
Evidence: Model profile ↗

131 · Gemini 3 ProNot specified991

Accuracy: 31.11%
USD/task: $0.81
Evidence: Model profile ↗

132 · Claude Haiku 5.5High990

Accuracy: 30.69%
USD/task: $0.05
Evidence: Model profile ↗

132 · Gemini 3.6 FlashLow990

Accuracy: 30.42%
USD/task: $0.22
Evidence: Model profile ↗

134 · GPT-5.6 LunaHigh987

Accuracy: 29.31%
USD/task: $0.29
Evidence: Model profile ↗

134 · GPT-5.4Low987

Accuracy: 29.17%
USD/task: $0.27
Evidence: Model profile ↗

136 · GLM 5.3 FlashLow984

Accuracy: 27.92%
USD/task: $0.06
Evidence: Model profile ↗

136 · Grok 4.6Low984

Accuracy: 27.64%
USD/task: $0.21
Evidence: Model profile ↗

138 · Muse Spark OpenMedium983

Accuracy: 27.5%
USD/task: $0.25
Evidence: Model profile ↗

139 · GPT-5.2Medium981

Accuracy: 26.67%
USD/task: $0.76
Evidence: Model profile ↗

140 · Claude Sonnet 5.5Low975

Accuracy: 24.03%
USD/task: $0.11
Evidence: Model profile ↗

141 · Beam 1.0xHigh974

Accuracy: 23.47%
USD/task: $0.61
Evidence: Model profile ↗

142 · Claude Opus 4.5Thinking, 16K973

Accuracy: 22.78%
USD/task: $0.79
Evidence: Model profile ↗

142 · Qwen3.8-27BLow973

Accuracy: 22.78%
USD/task: $0.24
Evidence: Model profile ↗

142 · GLM-5.2Not specified973

Accuracy: 22.78%
USD/task: $0.25
Evidence: Model profile ↗

145 · Beam 1.0High966

Accuracy: 20%
USD/task: $0.35
Evidence: Model profile ↗

146 · GPT-5.4 MinixHigh964

Accuracy: 18.9%
USD/task: $0.75
Evidence: Model profile ↗

146 · GPT-5.6 TerraLow964

Accuracy: 18.75%
USD/task: $0.17
Evidence: Model profile ↗

148 · GPT-5 ProNot specified963

Accuracy: 18.33%
USD/task: $7.14
Evidence: Model profile ↗

149 · GPT-6 LunaMedium962

Accuracy: 18.06%
USD/task: $0.01
Evidence: Model profile ↗

150 · GPT-5.1Thinking, High961

Accuracy: 17.64%
USD/task: $1.17
Evidence: Model profile ↗

151 · Grok 4.7Low959

Accuracy: 16.67%
USD/task: $0.17
Evidence: Model profile ↗

152 · Grok 4Thinking957

Accuracy: 15.98%
USD/task: $2.17
Evidence: Model profile ↗

153 · Claude Opus 4.5Thinking, 8K953

Accuracy: 13.89%
USD/task: $0.48
Evidence: Model profile ↗

154 · Inkling SmallMedium952

Accuracy: 13.61%
USD/task: $0.06
Evidence: Model profile ↗

154 · Claude Sonnet 4.5Thinking 32K952

Accuracy: 13.61%
USD/task: $0.76
Evidence: Model profile ↗

156 · Qwen3.8-27BMedium951

Accuracy: 13.19%
USD/task: $0.29
Evidence: Model profile ↗

156 · GPT-5.4 MiniHigh951

Accuracy: 13.19%
USD/task: $0.56
Evidence: Model profile ↗

158 · Gemini 3 Flash PreviewMedium950

Accuracy: 12.78%
USD/task: $0.08
Evidence: Model profile ↗

159 · Kimi K3Low949

Accuracy: 12.36%
USD/task: $0.25
Evidence: Model profile ↗

159 · Muse Spark OpenLow949

Accuracy: 12.22%
USD/task: $0.11
Evidence: Model profile ↗

161 · Kimi K2.5Not specified948

Accuracy: 11.81%
USD/task: $0.28
Evidence: Model profile ↗

162 · Gemini 3.5 Flash-LiteHigh945

Accuracy: 10.28%
USD/task: $0.14
Evidence: Model profile ↗

163 · GPT-5High944

Accuracy: 9.86%
USD/task: $0.73
Evidence: Model profile ↗

164 · GPT-5.2Low943

Accuracy: 9.72%
USD/task: $0.26
Evidence: Model profile ↗

164 · Claude Haiku 5.5Medium943

Accuracy: 9.44%
USD/task: $0.01
Evidence: Model profile ↗

166 · Gemini 3.5 FlashMinimal941

Accuracy: 8.89%
USD/task: $0.11
Evidence: Model profile ↗

166 · Claude Opus 4Thinking 16K941

Accuracy: 8.61%
USD/task: $1.93
Evidence: Model profile ↗

168 · Claude Opus 4.5Thinking, None939

Accuracy: 7.78%
USD/task: $0.22
Evidence: Model profile ↗

169 · GPT-5Medium938

Accuracy: 7.49%
USD/task: $0.45
Evidence: Model profile ↗

169 · GPT-5.6 Luna 2026-07-30Medium938

Accuracy: 7.36%
USD/task: $0.02
Evidence: Model profile ↗

169 · GPT-5.6 LunaMedium938

Accuracy: 7.36%
USD/task: $0.11
Evidence: Model profile ↗

172 · Claude Sonnet 4.5Thinking 16K937

Accuracy: 6.94%
USD/task: $0.35
Evidence: Model profile ↗

172 · Claude Sonnet 4.5Thinking 8K937

Accuracy: 6.94%
USD/task: $0.23
Evidence: Model profile ↗

174 · GPT-5.1Thinking, Medium936

Accuracy: 6.53%
USD/task: $0.42
Evidence: Model profile ↗

174 · o3High936

Accuracy: 6.53%
USD/task: $0.83
Evidence: Model profile ↗

174 · Beam 1.0Medium936

Accuracy: 6.39%
USD/task: $0.18
Evidence: Model profile ↗

177 · o4-miniHigh935

Accuracy: 6.11%
USD/task: $0.86
Evidence: Model profile ↗

177 · Claude Sonnet 4Thinking 16K935

Accuracy: 5.93%
USD/task: $0.49
Evidence: Model profile ↗

177 · Claude Sonnet 4.5Thinking 1K935

Accuracy: 5.83%
USD/task: $0.14
Evidence: Model profile ↗

180 · GPT-5.4 NanoxHigh934

Accuracy: 5.69%
USD/task: $0.16
Evidence: Model profile ↗

181 · Grok 4 (Fast Reasoning)Not specified933

Accuracy: 5.28%
USD/task: $0.06
Evidence: Model profile ↗

181 · Gemini 3.5 Flash-LiteMedium933

Accuracy: 5.28%
USD/task: $0.04
Evidence: Model profile ↗

181 · GPT-5.6 LunaLow933

Accuracy: 5.14%
USD/task: $0.07
Evidence: Model profile ↗

184 · Inkling SmallLow932

Accuracy: 4.86%
USD/task: $0.06
Evidence: Model profile ↗

184 · Gemini 2.5 ProThinking 32K932

Accuracy: 4.86%
USD/task: $0.76
Evidence: Model profile ↗

184 · GLM-5Not specified932

Accuracy: 4.86%
USD/task: $0.27
Evidence: Model profile ↗

184 · Minimax M2.5Not specified932

Accuracy: 4.86%
USD/task: $0.17
Evidence: Model profile ↗

184 · GPT-6 LunaLow932

Accuracy: 4.58%
USD/task: $0.01
Evidence: Model profile ↗

184 · Claude Opus 4Thinking 8K932

Accuracy: 4.52%
USD/task: $1.16
Evidence: Model profile ↗

190 · GPT-5.4 MiniMedium931

Accuracy: 4.44%
USD/task: $0.33
Evidence: Model profile ↗

190 · GPT-5 MiniHigh931

Accuracy: 4.44%
USD/task: $0.2
Evidence: Model profile ↗

190 · Claude Haiku 4.5Thinking 32K931

Accuracy: 4.03%
USD/task: $0.38
Evidence: Model profile ↗

190 · Deepseek V3.2Not specified931

Accuracy: 4.03%
USD/task: $0.12
Evidence: Model profile ↗

190 · Gemini 2.5 ProThinking 16K931

Accuracy: 4.03%
USD/task: $0.71
Evidence: Model profile ↗

190 · GPT-5 MiniMedium931

Accuracy: 4.03%
USD/task: $0.06
Evidence: Model profile ↗

190 · Inkling SmallMinimal931

Accuracy: 4.03%
USD/task: $0.05
Evidence: Model profile ↗

197 · Claude Sonnet 4.5Not specified930

Accuracy: 3.75%
USD/task: $0.13
Evidence: Model profile ↗

197 · GPT-5.6 Luna 2026-07-30Low930

Accuracy: 3.61%
USD/task: $0.01
Evidence: Model profile ↗

197 · GPT-5.4 NanoHigh930

Accuracy: 3.61%
USD/task: $0.13
Evidence: Model profile ↗

200 · Gemini 3 Flash PreviewMinimal929

Accuracy: 3.33%
USD/task: $0.02
Evidence: Model profile ↗

201 · o3-miniHigh928

Accuracy: 2.99%
USD/task: $0.55
Evidence: Model profile ↗

201 · o3Medium928

Accuracy: 2.98%
USD/task: $0.48
Evidence: Model profile ↗

201 · Gemini 2.5 ProThinking 8K928

Accuracy: 2.92%
USD/task: $0.44
Evidence: Model profile ↗

201 · Claude Haiku 4.5Thinking 16K928

Accuracy: 2.78%
USD/task: $0.14
Evidence: Model profile ↗

201 · Beam 1.0Low928

Accuracy: 2.78%
USD/task: $0.1
Evidence: Model profile ↗

201 · Claude Haiku 5.5Low928

Accuracy: 2.78%
USD/task: $0.01
Evidence: Model profile ↗

207 · Gemini 3.6 FlashMinimal927

Accuracy: 2.64%
USD/task: $0.07
Evidence: Model profile ↗

207 · GPT-5 NanoHigh927

Accuracy: 2.61%
USD/task: $0.03
Evidence: Model profile ↗

207 · Gemini 2.5 Flash (Preview)Thinking 24K927

Accuracy: 2.54%
USD/task: $0.32
Evidence: Model profile ↗

207 · o4-miniMedium927

Accuracy: 2.36%
USD/task: $0.23
Evidence: Model profile ↗

211 · Gemini 2.5 Flash (Preview)Thinking 1K926

Accuracy: 2.16%
USD/task: $0.03
Evidence: Model profile ↗

211 · Claude Sonnet 4Thinking 8K926

Accuracy: 2.12%
USD/task: $0.27
Evidence: Model profile ↗

211 · Gemini 2.5 Flash (Preview)Thinking 8K926

Accuracy: 2.12%
USD/task: $0.2
Evidence: Model profile ↗

211 · DeepSeek V4.1 Flash 1017None926

Accuracy: 2.08%
USD/task: $0.01
Evidence: Model profile ↗

211 · DeepSeek V4 Flash 0731None926

Accuracy: 2.08%
USD/task: $0
Evidence: Model profile ↗

211 · o3-miniMedium926

Accuracy: 2.08%
USD/task: $0.28
Evidence: Model profile ↗

211 · o3Low926

Accuracy: 1.99%
USD/task: $0.23
Evidence: Model profile ↗

211 · Gemini 2.5 Flash (Preview)Thinking 16K926

Accuracy: 1.98%
USD/task: $0.32
Evidence: Model profile ↗

211 · Nemotron 3.5 Super VLMedium926

Accuracy: 1.94%
USD/task: $0.02
Evidence: Model profile ↗

211 · GPT-5.1Thinking, Low926

Accuracy: 1.94%
USD/task: $0.13
Evidence: Model profile ↗

211 · GPT-5Low926

Accuracy: 1.94%
USD/task: $0.19
Evidence: Model profile ↗

211 · GPT-5.4 NanoMedium926

Accuracy: 1.94%
USD/task: $0.06
Evidence: Model profile ↗

223 · Gemini 2.5 Flash (Preview)Not specified925

Accuracy: 1.69%
USD/task: $0.06
Evidence: Model profile ↗

223 · Claude Haiku 4.5Thinking 8K925

Accuracy: 1.67%
USD/task: $0.09
Evidence: Model profile ↗

223 · GPT-5 MiniMinimal925

Accuracy: 1.67%
USD/task: $0.01
Evidence: Model profile ↗

223 · o4-miniLow925

Accuracy: 1.67%
USD/task: $0.05
Evidence: Model profile ↗

223 · GPT-6 SolNone925

Accuracy: 1.67%
USD/task: $0.07
Evidence: Model profile ↗

223 · GPT-5.4 NanoLow925

Accuracy: 1.53%
USD/task: $0.01
Evidence: Model profile ↗

223 · Qwen3.8-27BNone925

Accuracy: 1.53%
USD/task: $0.06
Evidence: Model profile ↗

223 · Gemini 3.5 Flash-LiteLow925

Accuracy: 1.53%
USD/task: $0.02
Evidence: Model profile ↗

223 · Nemotron 3.5 Super VLHigh925

Accuracy: 1.53%
USD/task: $0.04
Evidence: Model profile ↗

232 · Gemini 2.0 FlashNot specified924

Accuracy: 1.3%
USD/task: $0
Evidence: Model profile ↗

232 · Deepseek R1Not specified924

Accuracy: 1.3%
USD/task: $0.08
Evidence: Model profile ↗

232 · Claude Opus 4Not specified924

Accuracy: 1.27%
USD/task: $0.64
Evidence: Model profile ↗

232 · Claude Sonnet 4Not specified924

Accuracy: 1.27%
USD/task: $0.13
Evidence: Model profile ↗

232 · Codex Mini (Latest)Not specified924

Accuracy: 1.27%
USD/task: $0.23
Evidence: Model profile ↗

232 · Claude Haiku 4.5Not specified924

Accuracy: 1.25%
USD/task: $0.04
Evidence: Model profile ↗

232 · Claude Haiku 4.5Thinking 1K924

Accuracy: 1.25%
USD/task: $0.05
Evidence: Model profile ↗

232 · DeepSeek V4.1 FlashNone924

Accuracy: 1.25%
USD/task: $0.01
Evidence: Model profile ↗

232 · Gemini 3 Flash PreviewLow924

Accuracy: 1.25%
USD/task: $0.02
Evidence: Model profile ↗

232 · Qwen3-235b-a22b Instruct (25/07)Not specified924

Accuracy: 1.25%
USD/task: $0
Evidence: Model profile ↗

232 · Deepseek R1 (05/28)Not specified924

Accuracy: 1.12%
USD/task: $0.05
Evidence: Model profile ↗

232 · GPT-5.4 MiniLow924

Accuracy: 1.11%
USD/task: $0.06
Evidence: Model profile ↗

244 · Claude 3.78K923

Accuracy: 0.9%
USD/task: $0.36
Evidence: Model profile ↗

244 · GPT-5 NanoMedium923

Accuracy: 0.88%
USD/task: $0.01
Evidence: Model profile ↗

244 · Claude Sonnet 4Thinking 1K923

Accuracy: 0.85%
USD/task: $0.14
Evidence: Model profile ↗

244 · DeepSeek V4 Pro 0813None923

Accuracy: 0.83%
USD/task: $0.04
Evidence: Model profile ↗

244 · Gemini 3.5 Flash-LiteMinimal923

Accuracy: 0.83%
USD/task: $0.01
Evidence: Model profile ↗

244 · Nemotron 3.5 Super VLNone923

Accuracy: 0.83%
USD/task: $0.01
Evidence: Model profile ↗

244 · GPT-5.6 Luna 2026-07-30None923

Accuracy: 0.83%
USD/task: $0.01
Evidence: Model profile ↗

244 · GPT-5.2Not specified923

Accuracy: 0.83%
USD/task: $0.08
Evidence: Model profile ↗

244 · GPT-5 MiniLow923

Accuracy: 0.83%
USD/task: $0.02
Evidence: Model profile ↗

244 · o1-miniNot specified923

Accuracy: 0.83%
USD/task: $0.19
Evidence: Model profile ↗

244 · Gemini 1.5 ProNot specified923

Accuracy: 0.8%
USD/task: $0.04
Evidence: Model profile ↗

244 · GPT-4.5Not specified923

Accuracy: 0.8%
USD/task: $2.1
Evidence: Model profile ↗

244 · Claude 3.716K923

Accuracy: 0.7%
USD/task: $0.51
Evidence: Model profile ↗

257 · GPT-4.1Not specified922

Accuracy: 0.42%
USD/task: $0.07
Evidence: Model profile ↗

257 · GPT-5.1Thinking, None922

Accuracy: 0.42%
USD/task: $0.06
Evidence: Model profile ↗

257 · Grok 3 MiniLow922

Accuracy: 0.42%
USD/task: $0.01
Evidence: Model profile ↗

257 · Claude 3.71K922

Accuracy: 0.4%
USD/task: $0.14
Evidence: Model profile ↗

261 · Claude 3.7Not specified921

Accuracy: 0%
USD/task: $0.12
Evidence: Model profile ↗

261 · Claude Opus 4Thinking 1K921

Accuracy: 0%
USD/task: $0.75
Evidence: Model profile ↗

261 · Gemini 2.5 ProThinking 1K921

Accuracy: 0%
USD/task: $0.09
Evidence: Model profile ↗

261 · GPT-4.1-MiniNot specified921

Accuracy: 0%
USD/task: $0.01
Evidence: Model profile ↗

261 · GPT-4.1-NanoNot specified921

Accuracy: 0%
USD/task: $0
Evidence: Model profile ↗

261 · GPT-4oNot specified921

Accuracy: 0%
USD/task: $0.08
Evidence: Model profile ↗

261 · GPT-4o-miniNot specified921

Accuracy: 0%
USD/task: $0.01
Evidence: Model profile ↗

261 · GPT-5Minimal921

Accuracy: 0%
USD/task: $0.06
Evidence: Model profile ↗

261 · GPT-5 NanoLow921

Accuracy: 0%
USD/task: $0
Evidence: Model profile ↗

261 · GPT-5 NanoMinimal921

Accuracy: 0%
USD/task: $0
Evidence: Model profile ↗

261 · Grok 3Not specified921

Accuracy: 0%
USD/task: $0.14
Evidence: Model profile ↗

261 · Llama 4 MaverickNot specified921

Accuracy: 0%
USD/task: $0.01
Evidence: Model profile ↗

261 · Llama 4 ScoutNot specified921

Accuracy: 0%
USD/task: $0.01
Evidence: Model profile ↗

261 · Magistral MediumNot specified921

Accuracy: 0%
USD/task: $0.11
Evidence: Model profile ↗

261 · Magistral MediumThinking921

Accuracy: 0%
USD/task: $0.12
Evidence: Model profile ↗

261 · Magistral SmallNot specified921

Accuracy: 0%
USD/task: $0.05
Evidence: Model profile ↗

261 · o3-miniLow921

Accuracy: 0%
USD/task: $0.06
Evidence: Model profile ↗

261 · GPT-6 LunaNone921

Accuracy: 0%
USD/task: $0
Evidence: Model profile ↗

261 · Inkling SmallNone921

Accuracy: 0%
USD/task: $0.01
Evidence: Model profile ↗

Source & conversion

Verified foundation-model evaluations on the Semi-Private split. Competition systems, human baselines and unofficial or hidden results are excluded.

Accuracy and published USD per task.

How the Lumina conversion works

1,000 + 75 × ((original score − 34.93207487056949) / 33.36178799222151). The result is rounded once. Scores may tie.

Frozen initial verified cohort of 279 configurations; Each published configuration has one reference observation; no future automatic refit.

Calibration: arc-agi-2-z75-v1 · experimental
Reference mean: 34.93207487056949
Population SD: 33.36178799222151
Coverage: Contributes through the Lumina Reasoning Index.

Identity, exclusions & attribution

Verified publisher data, not an independent Lumina rerun.

Official leaderboard · Benchmark description

Publisher terms. Scores and resource figures are attributed to the original publisher.