Skip to content

Benchmarks

Leaderboard

All active models ranked by P50 latency — the median response time for a standard 500-token output, measured from EU (Amsterdam). Green < 500 ms, yellow 500–1000 ms, red > 1000 ms.

Filter:
#
1FLUX.1 Kontext [max] — Multi-Image FusionA0
2FLUX.1 Kontext [pro] — Multi-Image FusionB0
3gpt-5.6-terra0
4NVIDIA Nemotron Super 49B v1.5A16
5Qwen3-Coder-30B-A3B-InstructB81
6Mistral-Nemo-Instruct-2407C90
7SDXL 1.0C101
8gpt-5.2-chat-latestC121
9Mistral-Small-3.2-24B-Instruct-2506B121
10gpt-5.3-chat-latestC126
11Meta-Llama-3_3-70B-InstructB141
12Qwen2.5-VL-72B-InstructB150
13Mistral-7B-Instruct-v0.3C157
14Nous Hermes 3 70BA171
15Qwen 2.5 VL 72B InstructA175
16Llama 4 ScoutA181
17Mistral Voxtral Small 24BA209
18gpt-oss-20bC232
19Cohere Command-AA253
20Llama 3.3 70B InstructA273
21Gemini 2.5 FlashA345
22Qwen3.5-397B-A17BA378
23Gemini 2.5 Flash-LiteB401
24gpt-4.1-nanoC443
25gpt-oss-120bC447
26Qwen3-32BB463
27gpt-4.1-miniC524
28Qwen3.5-9BB527
29FLUX.1 SchnellB530
30gpt-5-nanoC535
31gpt-5.4-miniA537
32gpt-4o-miniC544
33o3C550
34o4-miniC570
35gpt-4oC571
36Llama 4 MaverickA576
37Qwen 3.6 PlusA581
38MiniMax M2.5A589
39gpt-5C608
40gpt-4.1B642
41Qwen 3.7 MaxA677
42o3-miniC704
43gpt-5.1B748
44gpt-5.4A786
45Claude Haiku 4.5A844
46gpt-5-miniC859
47gpt-5.5C933
48Claude Opus 4.7B940
49GLM-4.5V (vision)A953
50Qwen3.7 PlusB981
51gpt-5.2B990
52Claude Opus 4.8A1080
53Gemini 2.5 ProA1160
54gpt-5.4-nanoC1177
55Qwen3.7 MaxA1201
56Gemini 3.1 Flash LiteB1211
57Claude Sonnet 4.6A1236
58DeepSeek v4 ProA1317
59GLM-4.6V (vision)A1403
60Gemini Flash-Lite LatestC1431
61Claude Opus 4.5B1440
62Claude Opus 5A1528
63Claude Sonnet 4.5B1564
64gpt-4.1-nano-2025-04-14C1566
65gpt-5.4-mini-2026-03-17A1708
66DeepSeek v3.2A1723
67gpt-3.5-turboC1895
68gpt-3.5-turbo-1106C2018
69gpt-3.5-turbo-16kC2029
70gpt-4o-2024-08-06C2044
71Claude Opus 4.6B2072
72gpt-5.4-2026-03-05B2096
73GLM-5 TurboB2099
74gpt-4o-2024-11-20C2104
75gpt-5.1-2025-11-13B2201
76gpt-4.1-2025-04-14C2239
77gpt-5-search-api-2025-10-14B2290
78gpt-4o-2024-05-13C2294
79o3-2025-04-16B2335
80gpt-5.4-nano-2026-03-17A2454
81GLM-5A2544
82gpt-5.2-2025-12-11B2545
83gpt-3.5-turbo-0125C2562
84gpt-5.5-2026-04-23A2663
85Nano BananaB2873
86gpt-5-search-apiC2942
87o3-mini-2025-01-31C3085
88GLM-5.2A3120
89gpt-4o-mini-2024-07-18C3126
90Claude Sonnet 5A3189
91Gemini 3.5 FlashA3240
92gpt-4o-mini-search-previewC3253
93GLM-4.6A3598
94o4-mini-2025-04-16B3746
95Claude Fable 5A3982
96Gemini Flash LatestB4021
97Gemini Robotics-ER 1.6 PreviewB4190
98Nano Banana 2B4330
99gpt-4.1-mini-2025-04-14C4409
100Gemini 3 Flash PreviewC4415
101gpt-4o-search-previewC4750
102GLM-4.5 AirB4867
103gpt-5-nano-2025-08-07B5665
104o1-2024-12-17C5732
105gpt-4C6056
106Gemini 3.1 Pro PreviewC6084
107GLM-4.5A6314
108Gemini Pro LatestC6626
109o1C6680
110Gemini 3.1 Pro Preview Custom ToolsC7298
111CogView-4B7580
112gpt-5-mini-2025-08-07B7966
113gpt-4-turboC8019
114GLM-5.1A8184
115gpt-4-turbo-2024-04-09C8789
116gpt-4-0613C8877
117gpt-5-2025-08-07B9935
118Nano Banana ProA11201
119GLM-4.7A16423
120GLM ImageB30000

120 of 120 models · click column headers to sort

Fast (< 500 ms)
Medium (500–1000 ms)
Slow (> 1000 ms)
Updated every 6 hours · P50 = median latency · P95 = tail latency