ANALYSIS · LIVE FROM ARTIFICIAL ANALYSIS

Frontier model benchmarks.

Independent evaluations across the Artificial Analysis Intelligence Index, coding, math, MMLU-Pro, GPQA, LiveCodeBench, HLE, and price + throughput — refreshed every 30 minutes from artificialanalysis.ai.

576 models tracked · cached 30 min
AA INTELLIGENCE INDEX · TOP 10

Top by intelligence

01AClaude Fable 5 (Adaptive Reasoning, Max Effort, Opus 4.8 Fallback)Anthropic2026-06-09
59.9
02OGPT-5.6 Sol (max)OpenAI2026-07-09
58.9
03OGPT-5.6 Sol (xhigh)OpenAI2026-07-09
57.7
04KKimi K3Kimi2026-07-16
57.1
05OGPT-5.6 Sol (high)OpenAI2026-07-09
55.9
06AClaude Opus 4.8 (Adaptive Reasoning, Max Effort)Anthropic2026-05-28
55.7
07OGPT-5.6 Terra (max)OpenAI2026-07-09
55.0
08OGPT-5.5 (xhigh)OpenAI2026-04-23
54.8
09SGrok 4.5 (high)SpaceXAI2026-07-08
53.8
10OGPT-5.6 Sol (medium)OpenAI2026-07-09
53.6
AA CODING INDEX · TOP 10

Top by coding

01OGPT-5.6 Sol (xhigh)OpenAI2026-07-09
78.3
02OGPT-5.6 Sol (max)OpenAI2026-07-09
77.4
03OGPT-5.6 Sol (high)OpenAI2026-07-09
77.2
04OGPT-5.6 Terra (max)OpenAI2026-07-09
76.7
05AClaude Fable 5 (Adaptive Reasoning, Max Effort, Opus 4.8 Fallback)Anthropic2026-06-09
76.5
06OGPT-5.6 Sol (medium)OpenAI2026-07-09
76.3
07KKimi K3Kimi2026-07-16
76.2
08OGPT-5.5 (xhigh)OpenAI2026-04-23
74.9
09AClaude Opus 4.8 (Adaptive Reasoning, Max Effort)Anthropic2026-05-28
74.3
10AClaude Opus 4.7 (Adaptive Reasoning, Max Effort)Anthropic2026-04-16
73.6
AA MATH INDEX · TOP 10

Top by math

01OGPT-5.2 (xhigh)OpenAI2025-12-11
99.0
02OGPT-5 Codex (high)OpenAI2025-09-23
98.7
03GGemini 3 Flash Preview (Reasoning)Google2025-12-17
97.0
04OGPT-5.2 (medium)OpenAI2025-12-11
96.7
05DDeepSeek V3.2 SpecialeDeepSeek2025-12-01
96.7
06XMiMo-V2-Flash (Reasoning)Xiaomi2025-12-16
96.3
07OGPT-5.1 Codex (high)OpenAI2025-11-13
95.7
08GGemini 3 Pro Preview (high)Google2025-11-18
95.7
09ZGLM-4.7 (Reasoning)Z AI2025-12-22
95.0
10KKAT-Coder-Pro V1KwaiKAT2025-11-11
94.7
GPQA · TOP 10

GPQA (graduate-level Q&A)

01OGPT-5.6 Sol (max)OpenAI2026-07-09
0.9
02GGemini 3.1 Pro PreviewGoogle2026-02-19
0.9
03OGPT-5.5 (xhigh)OpenAI2026-04-23
0.9
04KKimi K3Kimi2026-07-16
0.9
05OGPT-5.5 (high)OpenAI2026-04-23
0.9
06OGPT-5.6 Sol (xhigh)OpenAI2026-07-09
0.9
07SGrok 4.5 (high)SpaceXAI2026-07-08
0.9
08MMiniMax-M3MiniMax2026-06-01
0.9
09OGPT-5.6 Sol (high)OpenAI2026-07-09
0.9
10OGPT-5.5 (medium)OpenAI2026-04-23
0.9
MMLU-PRO · TOP 10

MMLU-Pro

01GGemini 3 Pro Preview (high)Google2025-11-18
0.9
02GGemini 3 Pro Preview (low)Google2025-11-18
0.9
03AClaude Opus 4.5 (Reasoning)Anthropic2025-11-24
0.9
04GGemini 3 Flash Preview (Reasoning)Google2025-12-17
0.9
05AClaude Opus 4.5 (Non-reasoning)Anthropic2025-11-24
0.9
06GGemini 3 Flash Preview (Non-reasoning)Google2025-12-17
0.9
07AClaude 4.1 Opus (Reasoning)Anthropic2025-08-05
0.9
08AClaude 4.5 Sonnet (Reasoning)Anthropic2025-09-29
0.9
09MMiniMax-M2.1MiniMax2025-12-23
0.9
10OGPT-5.2 (xhigh)OpenAI2025-12-11
0.9
PRICING + THROUGHPUT

Cost-to-intelligence frontier

#Model$ / 1M (blended)ThroughputTTFT
01AClaude Fable 5 (Adaptive Reasoning, Max Effort, Opus 4.8 Fallback)Anthropic$20.0062 tok/s66.11s
02OGPT-5.6 Sol (max)OpenAI$11.2566 tok/s87.54s
03OGPT-5.6 Sol (xhigh)OpenAI$11.2559 tok/s30.82s
04KKimi K3Kimi$6.0059 tok/s1.35s
05OGPT-5.6 Sol (high)OpenAI$11.2553 tok/s9.93s
06AClaude Opus 4.8 (Adaptive Reasoning, Max Effort)Anthropic$10.0055 tok/s23.67s
07OGPT-5.6 Terra (max)OpenAI$5.63138 tok/s154.48s
08OGPT-5.5 (xhigh)OpenAI$11.2570 tok/s27.30s
09SGrok 4.5 (high)SpaceXAI$3.00119 tok/s12.24s
10OGPT-5.6 Sol (medium)OpenAI$11.2554 tok/s3.95s
11AClaude Opus 4.7 (Adaptive Reasoning, Max Effort)Anthropic$10.0049 tok/s22.95s
12AClaude Sonnet 5 (Adaptive Reasoning, Max Effort)Anthropic$4.0078 tok/s126.94s
13OGPT-5.5 (high)OpenAI$11.2564 tok/s13.07s
14OGPT-5.6 Terra (xhigh)OpenAI$5.63124 tok/s6.87s
15OGPT-5.4 (xhigh)OpenAI$5.63152 tok/s120.83s
← AIDB models indexSource · artificialanalysis.ai ↗