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:

💻 Coding (40%) — measuring programming ability, syntax accuracy, and algorithmic logic;
🧮 Math (15%) — testing quantitative problem-solving;
🔬 Reasoning (20%) — assessing logical thinking and multi-step deduction;
✍️ Instruction Following (25%) — evaluating how well the model follows custom constraints and user instructions.

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

Last updated: September 12, 2026

Rank	Model Name	Relative Quality	Score
1	Claude Fable 5.1	Overall Quality 95.08%	95.08
Coding 91.92% Math 99.06% Reasoning 96.55% Instr. Following 97.52%
2	GPT-6 Astra	Overall Quality 95.04%	95.04
Coding 96.92% Math — Reasoning 97.93% Instr. Following 90.91%
3	Claude Opus 5	Overall Quality 94.73%	94.73
Coding 94.09% Math 99.84% Reasoning 94.99% Instr. Following 97.36%
4	Claude Fable 5	Overall Quality 94.64%	94.64
Coding 98.21% Math 100.00% Reasoning 93.46% Instr. Following 91.32%
5	Kimi K3	Overall Quality 93.95%	93.95
Coding 95.78% Math 95.92% Reasoning 89.67% Instr. Following 95.04%
6	GPT-5.6 Sol	Overall Quality 93.63%	93.63
Coding 95.91% Math 94.03% Reasoning 92.00% Instr. Following 92.53%
7	GLM-5.3	Overall Quality 92.67%	92.67
Coding 94.40% Math 96.23% Reasoning 86.45% Instr. Following 94.71%
8	Gemini 3.8 Flash	Overall Quality 92.51%	92.51
Coding 93.26% Math 96.55% Reasoning 87.79% Instr. Following 96.36%
9	Gemini 3.7 Flash	Overall Quality 92.31%	92.31
Coding 91.47% Math 99.22% Reasoning 87.62% Instr. Following 95.04%
10	Muse Spark 1.3	Overall Quality 91.84%	91.84
Coding 90.44% Math — Reasoning 90.71% Instr. Following —
11	GPT-5.5	Overall Quality 91.00%	91.00
Coding 91.63% Math 95.92% Reasoning 86.74% Instr. Following 92.88%
12	Muse Spark 1.1	Overall Quality 90.48%	90.48
Coding 92.01% Math 93.72% Reasoning 86.75% Instr. Following 92.89%
13	Qwen3.8 Max	Overall Quality 90.34%	90.34
Coding 91.46% Math 95.92% Reasoning 86.58% Instr. Following 94.05%
14	Gemini 3.1 Pro	Overall Quality 89.76%	89.76
Coding 88.79% Math 94.03% Reasoning 85.86% Instr. Following 93.65%
15	Claude Opus 4.8	Overall Quality 89.64%	89.64
Coding 93.08% Math 94.98% Reasoning 89.17% Instr. Following 88.58%
16	GPT-5.6 Terra	Overall Quality 89.63%	89.63
Coding 92.68% Math 91.52% Reasoning 85.80% Instr. Following 88.91%
17	Claude Opus 4.7	Overall Quality 88.58%	88.58
Coding 93.84% Math 96.23% Reasoning 82.62% Instr. Following 88.59%
18	Grok 4.6	Overall Quality 88.46%	88.46
Coding 89.84% Math 86.34% Reasoning 86.56% Instr. Following 89.92%
19	Grok 4.5	Overall Quality 88.39%	88.39
Coding 90.00% Math 91.99% Reasoning 82.77% Instr. Following 92.07%
20	Claude Sonnet 5	Overall Quality 88.33%	88.33
Coding 90.60% Math 91.84% Reasoning 83.22% Instr. Following 91.90%
21	GLM-5.3-Flash	Overall Quality 88.22%	88.22
Coding 89.56% Math 97.33% Reasoning 83.44% Instr. Following 92.56%
22	Qwen3.7 Max	Overall Quality 88.05%	88.05
Coding 88.57% Math 94.19% Reasoning 82.67% Instr. Following 93.49%
23	GPT-5.4	Overall Quality 87.60%	87.60
Coding 88.03% Math 94.82% Reasoning 82.15% Instr. Following 92.18%
24	DeepSeek-V4-Pro-0813	Overall Quality 87.12%	87.12
Coding 87.45% Math 90.89% Reasoning 85.27% Instr. Following 89.98%
25	Qwen3.6 Max Preview	Overall Quality 86.99%	86.99
Coding 89.73% Math 91.84% Reasoning 71.79% Instr. Following 90.06%
26	Muse Spark 1.2	Overall Quality 86.96%	86.96
Coding 90.20% Math — Reasoning 76.75% Instr. Following 94.38%
27	Gemini 3.6 Flash	Overall Quality 86.78%	86.78
Coding 87.10% Math 96.55% Reasoning 79.66% Instr. Following 93.06%
28	Gemini 3.5 Flash	Overall Quality 86.65%	86.65
Coding 85.34% Math 95.76% Reasoning 82.19% Instr. Following 91.49%
29	GPT-5.6 Luna	Overall Quality 86.43%	86.43
Coding 88.08% Math 92.15% Reasoning 80.23% Instr. Following 88.43%
30	Claude Opus 4.6	Overall Quality 86.37%	86.37
Coding 91.12% Math 98.27% Reasoning 81.07% Instr. Following 87.32%
31	GLM-5.2	Overall Quality 85.99%	85.99
Coding 87.20% Math 92.46% Reasoning 80.95% Instr. Following 90.22%
32	Kimi K2.6	Overall Quality 85.36%	85.36
Coding 86.14% Math 92.31% Reasoning 79.02% Instr. Following 90.10%
33	GPT-5.2 Codex	Overall Quality 84.74%	84.74
Coding 84.56% Math 89.17% Reasoning 77.84% Instr. Following 91.29%
34	Muse Spark	Overall Quality 84.32%	84.32
Coding 84.51% Math 89.80% Reasoning 78.19% Instr. Following 91.26%
35	Grok-4.1	Overall Quality 84.01%	84.01
Coding 92.08% Math 86.81% Reasoning 47.41% Instr. Following 86.28%
36	Gemini 3 Pro	Overall Quality 83.82%	83.82
Coding 83.76% Math 92.15% Reasoning 77.60% Instr. Following 89.76%
37	MiMo-V2.5-Pro	Overall Quality 83.72%	83.72
Coding 85.51% Math 92.15% Reasoning 66.00% Instr. Following 93.64%
38	MiniMax M3	Overall Quality 83.44%	83.44
Coding 82.64% Math 85.40% Reasoning 79.35% Instr. Following 90.74%
39	Gemini 3 Flash	Overall Quality 83.18%	83.18
Coding 81.35% Math 91.99% Reasoning 75.14% Instr. Following 91.41%
40	Qwen3.7-Plus	Overall Quality 83.14%	83.14
Coding 83.24% Math 90.42% Reasoning 77.71% Instr. Following 90.14%
41	DeepSeek-V4-Flash-0731	Overall Quality 83.04%	83.04
Coding 81.93% Math 86.50% Reasoning 79.16% Instr. Following 89.05%
42	Claude Sonnet 4.5	Overall Quality 82.83%	82.83
Coding 85.22% Math 88.54% Reasoning 62.69% Instr. Following 91.74%
43	Qwen3.8-27B	Overall Quality 82.44%	82.44
Coding 85.00% Math 90.74% Reasoning 77.86% Instr. Following 85.12%
44	DeepSeek-V4.1-Flash	Overall Quality 82.40%	82.40
Coding 87.98% Math — Reasoning 78.01% Instr. Following —
45	Seed 2.0 Pro	Overall Quality 82.28%	82.28
Coding 84.64% Math 88.07% Reasoning 73.49% Instr. Following 86.94%
46	Claude Opus 4.5	Overall Quality 82.06%	82.06
Coding 87.20% Math 91.05% Reasoning 71.82% Instr. Following 85.85%
47	Qwen3.6 Plus	Overall Quality 81.95%	81.95
Coding 83.10% Math 88.70% Reasoning 71.41% Instr. Following 87.59%
48	GLM-5.1	Overall Quality 81.88%	81.88
Coding 83.47% Math 92.15% Reasoning 71.20% Instr. Following 90.79%
49	Claude Opus 4.1	Overall Quality 81.87%	81.87
Coding 84.40% Math 86.97% Reasoning 60.73% Instr. Following 90.91%
50	Qwen3.8-Flash-Next	Overall Quality 81.47%	81.47
Coding 78.13% Math — Reasoning 82.69% Instr. Following —
51	Grok 4.3	Overall Quality 81.39%	81.39
Coding 83.70% Math 83.20% Reasoning 66.72% Instr. Following 88.55%
52	GPT-5.3 Codex	Overall Quality 81.01%	81.01
Coding 73.09% Math — Reasoning 78.20% Instr. Following 88.73%
53	Claude Sonnet 4.6	Overall Quality 80.98%	80.98
Coding 85.83% Math 89.95% Reasoning 72.82% Instr. Following 84.62%
54	Hy3	Overall Quality 80.88%	80.88
Coding 82.89% Math 92.46% Reasoning 76.21% Instr. Following 83.45%
55	Gemma 4 31B	Overall Quality 80.72%	80.72
Coding 83.78% Math 91.52% Reasoning 64.78% Instr. Following 89.87%
56	Kimi K2.5	Overall Quality 80.50%	80.50
Coding 81.25% Math 91.21% Reasoning 71.96% Instr. Following 85.83%
57	MiMo-V2-Pro	Overall Quality 80.47%	80.47
Coding 82.58% Math 88.23% Reasoning 70.42% Instr. Following 85.99%
58	GPT-5.1 Thinking	Overall Quality 80.31%	80.31
Coding 80.57% Math 88.85% Reasoning 68.71% Instr. Following 88.37%
59	GLM-5	Overall Quality 80.14%	80.14
Coding 82.47% Math 86.66% Reasoning 68.10% Instr. Following 87.48%
60	Inkling	Overall Quality 79.92%	79.92
Coding 82.62% Math 91.99% Reasoning 71.28% Instr. Following 84.46%
61	Inkling-Small	Overall Quality 79.69%	79.69
Coding 82.34% Math 87.44% Reasoning 73.00% Instr. Following 80.00%
62	Muse Glimmer	Overall Quality 79.38%	79.38
Coding 85.91% Math 88.54% Reasoning 61.13% Instr. Following 83.64%
63	Qwen3.5-397B-A17B	Overall Quality 79.29%	79.29
Coding 78.61% Math 87.76% Reasoning 71.16% Instr. Following 89.17%
64	MiniMax M2.7	Overall Quality 78.77%	78.77
Coding 80.00% Math 83.52% Reasoning 68.90% Instr. Following 84.66%
65	Claude Opus 4	Overall Quality 78.42%	78.42
Coding 81.05% Math 83.52% Reasoning 57.64% Instr. Following 88.43%
66	GPT-5	Overall Quality 78.19%	78.19
Coding 78.97% Math 85.40% Reasoning 66.83% Instr. Following 84.44%
67	Gemma 4 26B-A4B	Overall Quality 77.99%	77.99
Coding 82.03% Math 90.58% Reasoning 57.34% Instr. Following 86.56%
68	MiMo-V2.5	Overall Quality 76.92%	76.92
Coding 79.79% Math 86.34% Reasoning 67.16% Instr. Following 83.64%
69	LongCat-Flash-Thinking-2601	Overall Quality 76.09%	76.09
Coding 78.83% Math 82.42% Reasoning 68.58% Instr. Following 79.17%
70	GLM-4.7	Overall Quality 75.79%	75.79
Coding 76.25% Math 84.62% Reasoning 66.66% Instr. Following 83.60%
71	MiMo-V2-Omni	Overall Quality 75.48%	75.48
Coding 78.02% Math 85.56% Reasoning 60.04% Instr. Following 83.17%
72	Qwen3 Max Thinking	Overall Quality 75.40%	75.40
Coding 74.02% Math 85.71% Reasoning 67.07% Instr. Following 84.75%
73	Qwen3.5-27B	Overall Quality 75.04%	75.04
Coding 72.60% Math 84.46% Reasoning 65.39% Instr. Following 83.85%
74	DeepSeek-V3.2	Overall Quality 75.02%	75.02
Coding 77.98% Math 84.62% Reasoning 64.37% Instr. Following 79.83%
75	MiniMax M2.5	Overall Quality 74.96%	74.96
Coding 76.10% Math 79.43% Reasoning 64.78% Instr. Following 79.85%
76	Claude Sonnet 4	Overall Quality 74.94%	74.94
Coding 78.03% Math 80.69% Reasoning 52.88% Instr. Following 83.64%
77	Gemini 3.1 Flash-Lite	Overall Quality 74.82%	74.82
Coding 76.29% Math 86.03% Reasoning 58.47% Instr. Following 85.08%
78	Kimi K2-Thinking-0905	Overall Quality 74.10%	74.10
Coding 75.50% Math 82.42% Reasoning 65.59% Instr. Following 79.20%
79	o3	Overall Quality 73.73%	73.73
Coding 73.57% Math 87.44% Reasoning 60.09% Instr. Following 81.97%
80	Qwen3.5-122B-A10B	Overall Quality 73.64%	73.64
Coding 72.31% Math 83.67% Reasoning 66.36% Instr. Following 84.66%
81	Gemini 3.5 Flash-Lite	Overall Quality 73.61%	73.61
Coding 78.25% Math 87.13% Reasoning 57.17% Instr. Following 88.10%
82	Step-3.5-Flash	Overall Quality 73.22%	73.22
Coding 73.96% Math 81.00% Reasoning 65.75% Instr. Following 78.31%
83	Kimi K2.7 Code	Overall Quality 73.04%	73.04
Coding 69.72% Math — Reasoning 72.25% Instr. Following 71.83%
84	MiniMax M2.1	Overall Quality 72.54%	72.54
Coding 72.82% Math 78.49% Reasoning 62.43% Instr. Following 79.35%
85	GLM-5V-Turbo	Overall Quality 72.33%	72.33
Coding 79.17% Math 86.81% Reasoning 45.05% Instr. Following 80.06%
86	MiMo-V2-Flash	Overall Quality 72.17%	72.17
Coding 72.02% Math 76.45% Reasoning 62.34% Instr. Following 80.08%
87	GPT-5 mini	Overall Quality 72.17%	72.17
Coding 74.06% Math 80.85% Reasoning 58.25% Instr. Following 80.19%
88	Grok-4	Overall Quality 71.82%	71.82
Coding 72.13% Math 84.14% Reasoning 66.63% Instr. Following 74.09%
89	GPT-4.5	Overall Quality 71.68%	71.68
Coding 71.87% Math 81.32% Reasoning 47.41% Instr. Following 87.27%
90	Qwen3-235B-A22B-Instruct-2507	Overall Quality 71.44%	71.44
Coding 72.66% Math 82.73% Reasoning 52.31% Instr. Following 83.64%
91	Mistral Medium 3.5	Overall Quality 71.37%	71.37
Coding 75.53% Math 84.62% Reasoning 52.23% Instr. Following 83.44%
92	Qwen3.5-35B-A3B	Overall Quality 71.06%	71.06
Coding 68.18% Math 80.06% Reasoning 61.92% Instr. Following 81.20%
93	ERNIE 5.0	Overall Quality 70.89%	70.89
Coding 83.05% Math 85.87% Reasoning 55.96% Instr. Following 72.07%
94	DeepSeek V3.1 Terminus	Overall Quality 70.74%	70.74
Coding 76.98% Math 79.59% Reasoning 53.05% Instr. Following 78.25%
95	Qwen3-Coder 480B A35B Instruct	Overall Quality 70.33%	70.33
Coding 74.45% Math 75.98% Reasoning 48.39% Instr. Following 78.51%
96	Grok-3 Mini	Overall Quality 70.23%	70.23
Coding 69.82% Math 78.18% Reasoning 62.13% Instr. Following 74.71%
97	DeepSeek-V3.2-Exp	Overall Quality 70.15%	70.15
Coding 75.11% Math 83.83% Reasoning 54.94% Instr. Following 76.41%
98	Gemini 2.5 Pro	Overall Quality 69.51%	69.51
Coding 70.98% Math 86.50% Reasoning 60.03% Instr. Following 76.84%
99	Grok 4 Fast	Overall Quality 68.92%	68.92
Coding 71.52% Math 83.99% Reasoning 60.27% Instr. Following 72.59%
100	GPT-5 Codex	Overall Quality 67.78%	67.78
Coding 53.58% Math — Reasoning 62.16% Instr. Following 87.32%