LLM Rankings: Top AI Models – Lewoniewski.info

This leaderboard presents the top 100 Large Language Models (LLMs) ranked by their Relative Quality Score. Unlike single-source lists, this ranking utilizes a unified, synthetic aggregation model to deliver a comprehensive and unbiased picture of current AI capabilities.

The ranking includes only those AI models that have been evaluated by at least two widely recognized independent benchmarking sources. Each model’s overall score is calculated as a composite across four key domains:

🤖 Agents (20%) — autonomous execution, environment interaction, and tool usage;
💻 Coding (30%) — measuring programming ability, syntax accuracy, and algorithmic logic;
🧮 Math (20%) — testing quantitative problem-solving;
🔬 Reasoning (30%) — assessing logical thinking and multi-step deduction.

Clicking on any model expands details across all four sub-aspects.

Last updated: September 26, 2026

Rank	Model Name	Relative Quality	Score
1	Claude Opus 5.5	Overall Quality 98.46%	98.46
Agents 92.76% Coding 99.88% Math 100.00% Reasoning 99.82%
2	Claude Fable 5.1	Overall Quality 93.88%	93.88
Agents 92.88% Coding 90.09% Math 97.45% Reasoning 95.97%
3	GPT-6 Astra	Overall Quality 93.22%	93.22
Agents 88.23% Coding 93.89% Math 90.65% Reasoning 97.58%
4	Claude Opus 5	Overall Quality 91.58%	91.58
Agents 86.55% Coding 87.82% Math 96.93% Reasoning 95.14%
5	Claude Fable 5	Overall Quality 91.55%	91.55
Agents 83.92% Coding 89.81% Math 96.23% Reasoning 95.24%
6	GPT-5.6 Sol	Overall Quality 90.15%	90.15
Agents 79.44% Coding 93.37% Math 90.52% Reasoning 93.82%
7	Muse Spark 1.3	Overall Quality 89.47%	89.47
Agents 79.00% Coding 87.62% Math 97.69% Reasoning 92.81%
8	Kimi K3	Overall Quality 88.39%	88.39
Agents 78.45% Coding 88.05% Math 92.01% Reasoning 92.94%
9	MiMo-V2.6-Pro	Overall Quality 87.18%	87.18
Agents 85.49% Coding 85.72% Math 94.30% Reasoning 85.03%
10	DeepSeek-V4.1-Flash	Overall Quality 86.47%	86.47
Agents 85.02% Coding 84.88% Math 90.87% Reasoning 86.09%
11	GPT-6 Sol	Overall Quality 86.39%	86.39
Agents 74.53% Coding 87.48% Math 92.06% Reasoning 89.44%
12	GLM-5.3	Overall Quality 86.25%	86.25
Agents 75.98% Coding 85.49% Math 92.73% Reasoning 89.56%
13	GPT-5.5	Overall Quality 86.22%	86.22
Agents 71.50% Coding 87.44% Math 92.61% Reasoning 90.56%
14	Claude Opus 4.8	Overall Quality 86.00%	86.00
Agents 75.17% Coding 82.57% Math 92.79% Reasoning 92.12%
15	Gemini 3.8 Flash	Overall Quality 85.81%	85.81
Agents 72.72% Coding 80.84% Math 98.01% Reasoning 91.38%
16	MiMo-V2.6-Flash	Overall Quality 85.33%	85.33
Agents 77.89% Coding 86.51% Math 91.46% Reasoning 85.01%
17	Qwen3.8	Overall Quality 85.17%	85.17
Agents 76.67% Coding 83.19% Math 88.69% Reasoning 90.45%
18	GPT-5.6 Terra	Overall Quality 84.94%	84.94
Agents 70.47% Coding 88.17% Math 87.47% Reasoning 89.68%
19	Gemini 3.7 Flash	Overall Quality 84.43%	84.43
Agents 66.33% Coding 81.78% Math 97.08% Reasoning 90.72%
20	Hy4	Overall Quality 84.08%	84.08
Agents 75.32% Coding 79.32% Math 90.52% Reasoning 90.38%
21	Claude Opus 4.7	Overall Quality 83.80%	83.80
Agents 70.01% Coding 81.79% Math 92.88% Reasoning 88.94%
22	DeepSeek-V4-Pro	Overall Quality 83.77%	83.77
Agents 73.00% Coding 80.55% Math 92.19% Reasoning 88.58%
23	Grok 4.6	Overall Quality 83.37%	83.37
Agents 70.91% Coding 81.72% Math 90.02% Reasoning 88.90%
24	Claude Sonnet 5	Overall Quality 83.22%	83.22
Agents 73.62% Coding 79.68% Math 90.92% Reasoning 88.04%
25	Muse Spark 1.1	Overall Quality 82.23%	82.23
Agents 64.05% Coding 79.68% Math 91.68% Reasoning 90.59%
26	DeepSeek-V4-Flash	Overall Quality 82.11%	82.11
Agents 80.28% Coding 77.23% Math 87.97% Reasoning 84.30%
27	Claude Opus 4.6	Overall Quality 82.10%	82.10
Agents 66.15% Coding 78.52% Math 93.65% Reasoning 88.60%
28	GLM-5.2	Overall Quality 81.73%	81.73
Agents 68.85% Coding 80.13% Math 92.68% Reasoning 84.63%
29	GLM-5.3-Flash	Overall Quality 81.52%	81.52
Agents 69.61% Coding 80.32% Math 88.70% Reasoning 85.86%
30	Grok 4.7	Overall Quality 81.30%	81.30
Agents 72.78% Coding 83.57% Math 88.97% Reasoning 79.60%
31	Grok 4.5	Overall Quality 81.21%	81.21
Agents 67.86% Coding 76.77% Math 92.35% Reasoning 87.11%
32	GPT-5.6 Luna	Overall Quality 80.43%	80.43
Agents 62.55% Coding 85.90% Math 83.36% Reasoning 84.94%
33	GPT-6 Luna	Overall Quality 79.87%	79.87
Agents 66.24% Coding 81.47% Math 87.44% Reasoning 82.31%
34	Muse Spark 1.2	Overall Quality 79.64%	79.64
Agents 60.41% Coding 77.05% Math 93.92% Reasoning 85.52%
35	Gemini 3.5 Flash	Overall Quality 79.16%	79.16
Agents 67.36% Coding 75.39% Math 86.09% Reasoning 86.16%
36	Qwen3.8-Flash-Next	Overall Quality 79.15%	79.15
Agents 69.78% Coding 77.35% Math 81.06% Reasoning 85.92%
37	Seed 2.1 Pro	Overall Quality 79.04%	79.04
Agents 75.65% Coding 67.23% Math 89.73% Reasoning 86.00%
38	GPT-5.4	Overall Quality 78.42%	78.42
Agents 63.54% Coding 70.91% Math 90.53% Reasoning 87.79%
39	Qwen3.7	Overall Quality 78.42%	78.42
Agents 54.52% Coding 76.33% Math 92.62% Reasoning 86.97%
40	Kimi K2.6	Overall Quality 78.34%	78.34
Agents 63.84% Coding 76.46% Math 88.20% Reasoning 83.30%
41	Gemini 3.1 Pro	Overall Quality 77.69%	77.69
Agents 49.10% Coding 74.77% Math 93.62% Reasoning 89.04%
42	Qwen3.8-27B	Overall Quality 77.41%	77.41
Agents 66.37% Coding 76.11% Math 83.09% Reasoning 82.27%
43	Gemini 3.6 Flash	Overall Quality 77.17%	77.17
Agents 50.43% Coding 75.45% Math 94.11% Reasoning 85.42%
44	Claude Sonnet 4.6	Overall Quality 75.69%	75.69
Agents 58.99% Coding 73.85% Math 86.16% Reasoning 81.69%
45	GPT-5.2	Overall Quality 74.63%	74.63
Agents 55.68% Coding 67.86% Math 89.14% Reasoning 84.36%
46	ChatGPT-4o	Overall Quality 74.29%	74.29
Agents — Coding 87.43% Math 81.17% Reasoning 72.77%
47	Qwen3.7-Plus	Overall Quality 74.18%	74.18
Agents 45.36% Coding 75.32% Math 89.72% Reasoning 81.87%
48	Muse Spark	Overall Quality 74.01%	74.01
Agents 45.42% Coding 73.39% Math 89.49% Reasoning 83.38%
49	Claude Opus 4.5	Overall Quality 73.93%	73.93
Agents 51.24% Coding 74.32% Math 86.10% Reasoning 80.54%
50	GLM-5.1	Overall Quality 73.92%	73.92
Agents 62.35% Coding 69.73% Math 85.22% Reasoning 78.30%
51	Grok Build 0.1	Overall Quality 73.68%	73.68
Agents 59.25% Coding 73.24% Math 80.74% Reasoning 79.02%
52	Qwen3.6 Plus	Overall Quality 73.43%	73.43
Agents 54.93% Coding 71.61% Math 87.94% Reasoning 77.90%
53	Hy3	Overall Quality 73.41%	73.41
Agents 49.85% Coding 71.92% Math 88.78% Reasoning 80.36%
54	Qwen3.6	Overall Quality 73.41%	73.41
Agents — Coding 69.05% Math 92.56% Reasoning 80.59%
55	ERNIE 5.0	Overall Quality 73.33%	73.33
Agents — Coding 90.87% Math 80.40% Reasoning 66.64%
56	Seed 2.0 Pro	Overall Quality 72.38%	72.38
Agents 49.74% Coding 71.40% Math 82.53% Reasoning 81.69%
57	Gemini 3 Pro	Overall Quality 71.27%	71.27
Agents 47.15% Coding 64.21% Math 88.46% Reasoning 82.94%
58	GPT-5.5 Instant	Overall Quality 71.04%	71.04
Agents — Coding 85.49% Math 68.79% Reasoning 72.11%
59	MiniMax M3	Overall Quality 70.92%	70.92
Agents 49.05% Coding 70.88% Math 76.62% Reasoning 81.75%
60	Gemini 3 Flash	Overall Quality 70.90%	70.90
Agents 51.47% Coding 62.49% Math 88.22% Reasoning 80.71%
61	MiMo-V2.5	Overall Quality 70.66%	70.66
Agents 57.51% Coding 65.35% Math 86.55% Reasoning 74.13%
62	GLM-5	Overall Quality 70.43%	70.43
Agents 54.58% Coding 64.30% Math 87.34% Reasoning 75.85%
63	Kimi K2.5	Overall Quality 70.36%	70.36
Agents 52.16% Coding 63.87% Math 87.67% Reasoning 77.45%
64	MiMo-V2-Pro	Overall Quality 70.18%	70.18
Agents 50.43% Coding 64.24% Math 88.61% Reasoning 77.01%
65	Qwen3.5-397B-A17B	Overall Quality 69.68%	69.68
Agents 55.27% Coding 60.93% Math 86.17% Reasoning 77.05%
66	Inkling	Overall Quality 69.20%	69.20
Agents 44.16% Coding 67.17% Math 84.24% Reasoning 77.90%
67	Kimi K2.7 Code	Overall Quality 68.82%	68.82
Agents 63.33% Coding 68.18% Math 64.19% Reasoning 76.20%
68	MiMo-V2.5-Pro	Overall Quality 68.74%	68.74
Agents 46.00% Coding 71.76% Math 79.46% Reasoning 73.73%
69	Kimi K2	Overall Quality 68.40%	68.40
Agents 48.36% Coding 67.78% Math 84.35% Reasoning 71.75%
70	LongCat-Flash	Overall Quality 68.27%	68.27
Agents 43.18% Coding 67.52% Math 80.00% Reasoning 77.91%
71	Qwen3 VL 235B A22B	Overall Quality 68.09%	68.09
Agents 43.35% Coding 86.83% Math 77.24% Reasoning 59.73%
72	GPT-5.4 mini	Overall Quality 67.96%	67.96
Agents 54.00% Coding 66.13% Math 74.80% Reasoning 74.55%
73	MiniMax M2.7	Overall Quality 67.95%	67.95
Agents 48.01% Coding 64.47% Math 84.02% Reasoning 74.00%
74	GPT-5.3 Codex	Overall Quality 67.93%	67.93
Agents 59.93% Coding 70.03% Math 58.93% Reasoning 77.17%
75	INTELLECT-3	Overall Quality 67.31%	67.31
Agents — Coding 79.79% Math 76.27% Reasoning 60.41%
76	GPT-5.2 Codex	Overall Quality 67.31%	67.31
Agents 58.81% Coding 61.59% Math 72.40% Reasoning 75.30%
77	GPT-5.4 nano	Overall Quality 67.26%	67.26
Agents 50.05% Coding 67.68% Math 76.43% Reasoning 72.20%
78	Inkling-Small	Overall Quality 67.20%	67.20
Agents 39.79% Coding 65.38% Math 83.64% Reasoning 76.35%
79	GPT-5.1	Overall Quality 67.04%	67.04
Agents — Coding 59.25% Math 82.61% Reasoning 75.81%
80	Qwen3.6-27B	Overall Quality 66.99%	66.99
Agents 49.69% Coding 68.29% Math 80.61% Reasoning 68.15%
81	Muse Glimmer	Overall Quality 66.88%	66.88
Agents — Coding 58.57% Math 88.13% Reasoning 72.26%
82	DeepSeek V3.1 Terminus	Overall Quality 66.81%	66.81
Agents — Coding 69.02% Math 80.06% Reasoning 66.97%
83	Step-3.5-Flash	Overall Quality 66.69%	66.69
Agents 47.15% Coding 66.28% Math 80.55% Reasoning 70.87%
84	Gemini 3.5 Flash-Lite	Overall Quality 66.43%	66.43
Agents 52.88% Coding 63.48% Math 83.21% Reasoning 67.22%
85	Grok-4	Overall Quality 66.20%	66.20
Agents — Coding 62.34% Math 79.92% Reasoning 71.72%
86	MiMo-V2-Omni	Overall Quality 65.78%	65.78
Agents 36.61% Coding 68.42% Math 86.08% Reasoning 69.04%
87	Mistral	Overall Quality 65.71%	65.71
Agents — Coding 85.48% Math 79.91% Reasoning 46.94%
88	Qwen3	Overall Quality 65.18%	65.18
Agents 38.17% Coding 63.98% Math 81.11% Reasoning 73.76%
89	MiniMax M2.5	Overall Quality 65.16%	65.16
Agents 50.26% Coding 60.47% Math 79.43% Reasoning 70.28%
90	Gemma 4 31B	Overall Quality 65.15%	65.15
Agents 44.21% Coding 59.95% Math 81.83% Reasoning 73.19%
91	Claude Sonnet 4.5	Overall Quality 64.92%	64.92
Agents 50.43% Coding 61.00% Math 77.24% Reasoning 70.27%
92	Grok 4.3	Overall Quality 64.69%	64.69
Agents 31.31% Coding 65.15% Math 84.64% Reasoning 73.20%
93	GLM-4.5V	Overall Quality 64.58%	64.58
Agents — Coding 77.69% Math 74.21% Reasoning 54.76%
94	Qwen3.5-122B-A10B	Overall Quality 64.53%	64.53
Agents 47.50% Coding 55.48% Math 83.72% Reasoning 72.14%
95	Step 5	Overall Quality 64.47%	64.47
Agents — Coding 73.22% Math — Reasoning 75.00%
96	GPT-5	Overall Quality 64.34%	64.34
Agents 36.44% Coding 62.63% Math 80.72% Reasoning 73.72%
97	GPT-5 mini	Overall Quality 64.27%	64.27
Agents — Coding 67.68% Math 73.76% Reasoning 64.05%
98	Qwen2.5	Overall Quality 63.90%	63.90
Agents — Coding 77.69% Math 74.68% Reasoning 52.18%
99	GLM-4.7	Overall Quality 63.77%	63.77
Agents 37.13% Coding 58.71% Math 82.61% Reasoning 74.01%
100	Ling 3.0 Flash	Overall Quality 63.45%	63.45
Agents 54.23% Coding 61.11% Math 72.83% Reasoning 65.68%