Benchmarks / Cost

AA Intelligence Index Cost per Task

Measured USD per task, weighted across the Intelligence Index evaluations.

112models
–Open Index weight
–last updated

AA Intelligence Index Cost per Task

Lower is better · best effort per model

Chart settings
112 of 112 selected

full table below

#ModelEffortScoreCost / taskHarness
1GPT-6 Lunalow0.00 USD––
2Granite 4.2 3B–0.01 USD––
3Ministral 3 3Bnone0.01 USD––
4Gemma 4 E4B (Reasoning)–0.01 USD––
5GPT-5.6 Lunalow0.01 USD––
6Ministral 3 8Bnone0.01 USD––
7gpt-oss-20bhigh0.01 USD––
8Mistral Small 4–0.02 USD––
9NVIDIA Nemotron 3 Nano 30B A3B (Reasoning)–0.02 USD––
10MiMo V2.5–0.02 USD––
11Ministral 3 14Bnone0.02 USD––
12DeepSeek V3none0.02 USD––
13Granite 4.2 8B–0.02 USD––
14Haiku 5.5low0.02 USD––
15Mistral Small 3.1none0.03 USD––
16Mistral Large 3none0.03 USD––
17Llama 4 Mavericknone0.04 USD––
18Mercury 2high0.04 USD––
19Celeris-1none0.05 USD––
20GPT-5 Minihigh0.05 USD––
21Muse Glimmer (High)high0.05 USD––
22MiMo V2.5 Pro–0.05 USD––
23Gemini 3.1 Flash-Litehigh0.06 USD––
24MiMo V2.6 Flash–0.06 USD––
25Qwen 3 30B-A3B–0.07 USD––
26Granite 4.2 30B–0.07 USD––
27Hy3–0.07 USD––
28Solar Pro 3high0.08 USD––
29Inkling Small–0.09 USD––
30DeepSeek V4 Flash 0420max0.09 USD––
31Nemotron 3.5 Lightning–0.09 USD––
32gpt-oss-120bhigh0.11 USD––
33Mistral Small 3.2none0.11 USD––
34Llama 4 Scoutnone0.11 USD––
35Mercury 2.5–0.12 USD––
36Trinity Large Thinking–0.12 USD––
37DeepSeek V4 Pro 0424max0.12 USD––
38GPT-6.1 Sollow0.13 USD––
39GPT-6 Sollow0.13 USD––
40MiMo V2.6 Pro–0.13 USD––
41GPT-5.6 Terranone0.14 USD––
42LongCat 2.0–0.15 USD––
43DeepSeek V4.1 Flashnone0.15 USD––
44Grok 4.3none0.17 USD––
45GPT-5.4 Nanoxhigh0.18 USD––
46Gemini 3.5 Flash-Litehigh0.19 USD––
47Qwen 3.5 9B–0.21 USD––
48Qwen 3.7 Plus–0.22 USD––
49DeepSeek V4 Flash 0731max0.22 USD––
50Nemotron 3.5 Super VL–0.25 USD––
51GPT-5.6 Sollow0.26 USD––
52DeepSeek R1–0.26 USD––
53Haiku 4.5–0.28 USD––
54GLM 5.3 Flashmax0.28 USD––
55Ring-2.6-1Txhigh0.29 USD––
56DeepSeek V4 Flashmax0.31 USD––
57Qwen 3.5 122B-A10B–0.32 USD––
58Gemini 2.5 Pro–0.33 USD––
59Sonnet 5.5low0.35 USD––
60DeepSeek V4 Pro 0813none0.36 USD––
61Solar Mini 4–0.37 USD––
62Qwen 3.8 Flash–0.37 USD––
63Grok 4.6low0.38 USD––
64GPT-5.4 Minixhigh0.45 USD––
65Apodex 1.1–0.46 USD––
66Qwen 3.5 397B-A17B–0.47 USD––
67Qwen 3.6 35B-A3B–0.48 USD––
68Mistral Medium 3.5high0.50 USD––
69MiniMax M3–0.51 USD––
70Sonnet 5low0.51 USD––
71Kimi K2.7 Code–0.54 USD––
72Opus 5.5low0.55 USD––
73Sonnet 4.5–0.56 USD––
74Nemotron 3 Ultramax0.60 USD––
75Qwen 3.6 27B–0.62 USD––
76GPT-5.5 Instant–0.69 USD––
77Solar Pro 4–0.71 USD––
78Qwen 3 Codernone0.78 USD––
79Kimi K2.6–0.80 USD––
80GPT-6 Astralow0.82 USD––
81Muse Spark Open (Xhigh)xhigh0.84 USD––
82GLM 5.3low0.85 USD––
83Grok 4.5high0.88 USD––
84GPT-5.5medium0.90 USD––
85Gemini 3.7 Flashhigh0.93 USD––
86Gemini 3.8 Flashmedium0.93 USD––
87Muse Spark 1.2xhigh0.97 USD––
88Ling 3.1 Flash–1.00 USD––
89Qwen 3.8 27Bxhigh1.01 USD––
90GLM 5.1–1.02 USD––
91Step 5 Preview–1.03 USD––
92Opus 5low1.10 USD––
93Mistral Large 4–1.13 USD––
94Kimi K3low1.15 USD––
95Qwen 3.7 Maxmax1.15 USD––
96Grok 4.7low1.25 USD––
97Gemini 3.1 Pro–1.30 USD––
98Muse Spark 1.3xhigh1.37 USD––
99Muse Spark 1.1xhigh1.38 USD––
100GLM 5.2max1.47 USD––
101Gemini 3.6 Flashhigh1.60 USD––
102Nemotron 3 Super 120B A12B (Reasoning)–1.64 USD––
103Gemini 4high1.99 USD––
104Quasar 438B (Max, Based on GLM-5.2)max2.02 USD––
105Qwen 3.8 2.4T-A95B–2.16 USD––
106Fable 5.1low2.37 USD––
107Sonnet 4.6max2.49 USD––
108Qwen 3.8 Maxmax2.67 USD––
109Gemini 3.5 Flashhigh3.69 USD––
110Opus 4.8max4.08 USD––
111Muse Wondermax4.52 USD––
112Fable 5max8.75 USD––