AI Model Rankings

Measured success rate and latency for the models we serve, from real API calls made through ApexApi during health sweeps over the last 30 days. These are our own measured numbers, not vendor claims.

#	Model	Checks	Success	Avg latency	Last checked
1	Mistral Large 3 2512 bedrock	2	100%	2.2s	Jul 13
2	GPT-OSS 20B bedrock	2	100%	2.5s	Jul 13
3	o3 openai	3	100%	2.5s	Jul 13
4	GPT-5 Chat openai	3	100%	2.5s	Jul 13
5	Qwen3 VL Flash alibaba	3	100%	2.6s	Jul 13
6	Qwen3 VL Plus alibaba	3	100%	2.7s	Jul 13
7	GPT-4.1 openai	3	100%	2.7s	Jul 13
8	Gemini 2.5 Flash Lite google	3	100%	2.9s	Jul 13
9	Nano Banana google	3	100%	2.9s	Jul 13
10	GPT-OSS 120B bedrock	2	100%	3.0s	Jul 13
11	Mistral Medium 3.5 mistral	3	100%	3.0s	Jul 13
12	Gemini 3 Flash Preview google	3	100%	3.0s	Jul 13
13	Codestral 2508 mistral	3	100%	3.0s	Jul 13
14	Claude Opus 4.8 anthropic	3	100%	3.1s	Jul 13
15	GPT-4o openai	3	100%	3.1s	Jul 13
16	Ministral 3 3B 2512 mistral	3	100%	3.1s	Jul 13
17	GPT-4o openai	3	100%	3.1s	Jul 13
18	Claude Haiku 4.5 anthropic-vertex	2	100%	3.1s	Jul 13
19	MiniMax M2.5 bedrock	2	100%	3.2s	Jul 13
20	Gemini 3.1 Flash Lite google	3	100%	3.2s	Jul 13
21	Mistral Medium 3 mistral	3	100%	3.2s	Jul 13
22	o3 Mini openai	3	100%	3.3s	Jul 13
23	o1 openai	3	100%	3.3s	Jul 13
24	Ministral 3 3B 2512 bedrock	2	100%	3.3s	Jul 13
25	GPT-5.4 Mini openai	3	100%	3.4s	Jul 13
26	Gemini 3.1 Flash Lite Preview google	3	100%	3.4s	Jul 13
27	Claude Opus 4.7 anthropic	3	100%	3.4s	Jul 13
28	Claude Opus 4.5 anthropic-vertex	2	100%	3.4s	Jul 13
29	GPT-5.4 openai	3	100%	3.4s	Jul 13
30	GLM 4.7 Flash bedrock	2	100%	3.5s	Jul 13
31	GPT-4o openai	3	100%	3.5s	Jul 13
32	Claude Opus 4.7 anthropic-vertex	2	100%	3.5s	Jul 13
33	Ministral 3 14B 2512 mistral	3	100%	3.5s	Jul 13
34	GPT-4o openai	3	100%	3.5s	Jul 13
35	GPT-5.6 Sol openai	1	100%	3.5s	Jul 13
36	Mistral Small 4 mistral	3	100%	3.6s	Jul 13
37	Claude Opus 4.8 anthropic-vertex	2	100%	3.6s	Jul 13
38	GPT-5.1 openai	3	100%	3.6s	Jul 13
39	Devstral 2 2512 mistral	3	100%	3.6s	Jul 13
40	Qwen3 Coder Plus alibaba	3	100%	3.6s	Jul 13
41	GPT-4o-mini openai	3	100%	3.7s	Jul 13
42	GPT-4.1 Nano openai	3	100%	3.7s	Jul 13
43	GPT-5.4 Nano openai	3	100%	3.7s	Jul 13
44	GPT-5.2 openai	3	100%	3.8s	Jul 13
45	Kimi K2.5 bedrock	2	100%	3.8s	Jul 13
46	GPT-5.6 Luna openai	1	100%	3.8s	Jul 13
47	Kimi K2 Thinking bedrock	2	100%	3.8s	Jul 13
48	Ministral 3 14B 2512 bedrock	2	100%	3.8s	Jul 13
49	GPT-4 Turbo openai	3	100%	3.8s	Jul 13
50	GPT-4o-mini openai	3	100%	3.8s	Jul 13
51	Claude Haiku 4.5 anthropic	3	100%	3.8s	Jul 13
52	GPT-5.5 openai	3	100%	3.9s	Jul 13
53	o4 Mini openai	3	100%	3.9s	Jul 13
54	Devstral 2 2512 bedrock	2	100%	3.9s	Jul 13
55	Claude Sonnet 4.6 anthropic	3	100%	4.1s	Jul 13
56	GPT-5.6 Terra openai	1	100%	4.1s	Jul 13
57	Claude Opus 4.6 anthropic	3	100%	4.1s	Jul 13
58	Claude Opus 4.6 anthropic-vertex	2	100%	4.1s	Jul 13
59	Gemini 3.5 Flash google	3	100%	4.2s	Jul 13
60	Ministral 3 8B 2512 bedrock	2	100%	4.2s	Jul 13
61	GLM 4.7 bedrock	2	100%	4.3s	Jul 13
62	GPT-4 openai	3	100%	4.3s	Jul 13
63	Mistral Large 3 2512 mistral	3	100%	4.4s	Jul 13
64	Gemini 2.5 Flash google	3	100%	4.5s	Jul 13
65	GPT-5 Mini openai	3	100%	4.6s	Jul 13
66	GPT-3.5 Turbo 16k openai	3	100%	4.6s	Jul 13
67	DeepSeek V4 Pro deepseek	3	100%	4.7s	Jul 13
68	Qwen3 Max alibaba	3	100%	4.7s	Jul 13
69	Qwen3.6 Flash alibaba	3	100%	4.7s	Jul 13
70	Gemma 4 31B google	3	100%	4.7s	Jul 13
71	GPT-4.1 Mini openai	3	100%	4.8s	Jul 13
72	GPT-5 openai	3	100%	4.8s	Jul 13
73	Claude Sonnet 4.6 anthropic-vertex	2	100%	4.9s	Jul 13
74	GPT-3.5 Turbo openai	3	100%	4.9s	Jul 13
75	GPT-4o Search Preview openai	3	100%	5.0s	Jul 13
76	Ministral 3 8B 2512 mistral	3	100%	5.0s	Jul 13
77	Claude Opus 4.5 anthropic	3	100%	5.3s	Jul 13
78	DeepSeek V4 Flash deepseek	3	100%	5.8s	Jul 13
79	Gemini 3.1 Pro Preview google	3	100%	6.0s	Jul 13
80	Gemini 3.1 Pro Preview Custom Tools google	3	100%	6.1s	Jul 13
81	GPT-5 Nano openai	3	100%	6.2s	Jul 13
82	Gemma 4 26B A4B google	3	100%	6.4s	Jul 13
83	Qwen3.5 Flash alibaba	3	100%	7.0s	Jul 13
84	Qwen3.7 Max alibaba	3	100%	7.5s	Jul 13
85	GPT-4o-mini Search Preview openai	3	100%	7.6s	Jul 13
86	Qwen3.7 Plus alibaba	3	100%	9.5s	Jul 13
87	Gemini 2.5 Pro google	3	100%	9.6s	Jul 13
88	GLM 5 bedrock	2	100%	22.3s	Jul 13
89	Qwen3.5 Plus alibaba	3	67%	11.1s	Jul 13
90	Pixtral Large 2411 mistral	3	33%	4.8s	Jul 13

Latency is the average of successful calls and depends on prompt size, region, and upstream load. Success rate reflects our smoke tests, not a guarantee. Call any of these through one API and key at the model catalog.