AI Model Arena

43 frontier AI models scored across 8 dimensions — reasoning, coding, math, multimodal, speed, cost, context, safety. Side-by-side comparison and dynamic ranking. Updated April 2026.

Score Weights:Reasoning 20%Coding 20%Math 15%Multimodal 15%Speed 10%Cost Efficiency 10%Context 5%Safety 5%

42 models tracked · Auto-generated badges show the leader in each category.

#	Model	Company	Total	Reasoning	Coding	Math	Multimodal	Speed	Cost Efficiency	Context	Safety
🥇	Gemini 3.1 Pro🇺🇸Best OverallBest Multimodal Google DeepMind · 2026-04	Google DeepMind 2026-04	92.6	96	93	96	99	84	76	100	92
🥈	DeepSeek V4🇨🇳Best Value DeepSeek · 2026-04	DeepSeek 2026-04	91.6	94	96	96	78	90	99	88	84
🥉	Gemini 2.5 Pro🇺🇸 Google DeepMind · 2026-01	Google DeepMind 2026-01	91.3	94	91	95	97	82	78	100	90
4	GPT-5.5🇺🇸Best Reasoning OpenAI · 2026-04	OpenAI 2026-04	91.1	98	95	97	96	78	65	92	93
5	Claude Sonnet 4.6🇺🇸 Anthropic · 2026-01	Anthropic 2026-01	91.1	92	95	89	86	92	88	95	94
6	Claude Opus 4.7🇺🇸Best CodingSafest Anthropic · 2026-03	Anthropic 2026-03	90.3	96	97	93	89	78	70	95	96
7	Qwen3.5 Max🇨🇳 Alibaba · 2026-03	Alibaba 2026-03	90.1	92	91	92	89	84	91	95	82
8	GPT-5🇺🇸 OpenAI · 2025-12	OpenAI 2025-12	90.1	97	94	96	95	75	68	88	92
9	Llama 4.5🇺🇸 Meta · 2026-03	Meta 2026-03	89.7	91	90	88	92	85	93	90	84
10	DeepSeek R2🇨🇳 DeepSeek · 2026-02	DeepSeek 2026-02	89.4	96	93	97	76	75	96	88	82
11	GPT-4.1🇺🇸 OpenAI · 2025-04	OpenAI 2025-04	89.1	90	92	88	92	86	80	92	91
12	Grok 5🇺🇸 xAI · 2026-03	xAI 2026-03	88.6	95	91	94	90	86	70	88	76
13	DeepSeek V3.5🇨🇳 DeepSeek · 2026-02	DeepSeek 2026-02	88.4	91	93	94	70	88	98	85	82
14	Claude 3.7 Sonnet🇺🇸 Anthropic · 2025-02	Anthropic 2025-02	88.2	88	92	86	84	88	86	92	94
15	Qwen3 Max🇨🇳 Alibaba · 2025-09	Alibaba 2025-09	87.8	89	88	90	86	84	90	92	80
16	Gemini 2.0 Flash🇺🇸Fastest Google DeepMind · 2025-02	Google DeepMind 2025-02	87.5	84	82	85	90	96	95	90	88
17	GLM-5🇨🇳 Zhipu AI · 2026-02	Zhipu AI 2026-02	87.4	89	87	89	84	86	91	88	82
18	Mistral Large 3🇫🇷 Mistral AI · 2026-02	Mistral AI 2026-02	87.2	89	90	87	84	86	84	88	86
19	o4🇺🇸Best Math OpenAI · 2026-02	OpenAI 2026-02	87.0	98	94	99	90	58	55	84	94
20	Claude Haiku 4.5🇺🇸 Anthropic · 2025-10	Anthropic 2025-10	87.0	86	88	82	80	95	92	90	93
21	Kimi K2.5🇨🇳 Moonshot AI · 2026-03	Moonshot AI 2026-03	86.9	89	87	86	84	84	88	98	82
22	GPT-4o🇺🇸 OpenAI · 2024-05	OpenAI 2024-05	86.8	86	86	84	90	92	86	80	90
23	Grok 4🇺🇸 xAI · 2025-11	xAI 2025-11	86.7	93	88	92	87	85	72	80	78
24	Llama 4🇺🇸 Meta · 2025-04	Meta 2025-04	86.7	88	86	84	88	84	92	88	82
25	Gemini 1.5 Pro🇺🇸 Google DeepMind · 2024-05	Google DeepMind 2024-05	85.9	86	84	86	92	78	82	95	88
26	o3🇺🇸 OpenAI · 2025-04	OpenAI 2025-04	85.7	96	92	98	88	55	60	80	93
27	o3-mini🇺🇸 OpenAI · 2025-01	OpenAI 2025-01	85.6	91	89	93	70	78	88	80	90
28	DeepSeek R1🇨🇳 DeepSeek · 2025-01	DeepSeek 2025-01	85.4	94	90	96	65	70	96	80	78
29	GLM-4.5🇨🇳 Zhipu AI · 2025-07	Zhipu AI 2025-07	85.1	86	84	87	80	86	92	84	80
30	Doubao Pro🇨🇳 ByteDance · 2024-12	ByteDance 2024-12	84.3	82	80	82	86	90	96	82	80
31	Kimi K2🇨🇳 Moonshot AI · 2025-08	Moonshot AI 2025-08	84.3	86	85	84	78	82	88	96	80
32	ERNIE 5🇨🇳 Baidu · 2025-11	Baidu 2025-11	84.1	84	82	84	86	84	88	82	82
33	Yi-Lightning🇨🇳 01.AI · 2024-10	01.AI 2024-10	83.3	84	80	82	76	94	95	78	80
34	Mistral Large 2🇫🇷 Mistral AI · 2024-07	Mistral AI 2024-07	82.8	84	86	82	75	86	84	80	85
35	Amazon Nova Pro🇺🇸 Amazon · 2024-12	Amazon 2024-12	82.5	80	78	80	84	88	90	84	86
36	MiniMax Text-01🇨🇳 MiniMax · 2025-01	MiniMax 2025-01	82.1	82	80	82	78	80	90	96	78
37	Phi-4🇺🇸 Microsoft · 2024-12	Microsoft 2024-12	80.4	80	78	88	60	92	96	70	86
38	Sonar Large🇺🇸 Perplexity · 2025-01	Perplexity 2025-01	80.1	82	76	78	72	90	88	80	84
39	Command R+🇨🇦 Cohere · 2024-08	Cohere 2024-08	79.8	82	78	78	70	86	84	84	88
40	Qwen2.5-Coder 32B🇨🇳 Alibaba · 2024-11	Alibaba 2024-11	79.8	80	90	82	50	86	95	80	78
41	Llama 3.3 70B🇺🇸 Meta · 2024-12	Meta 2024-12	79.5	82	82	80	65	80	90	78	80
42	Codestral 2🇫🇷 Mistral AI · 2025-01	Mistral AI 2025-01	79.0	78	88	80	50	90	92	82	80