Benchmarks / Speed

Provider latency

Median provider latency across OpenRouter endpoints, in seconds (3d).

179models
–Open Index weight
–last updated
OpenRouterpublisher

Provider latency

Lower is better · best effort per model

Chart settings
179 of 179 selected

full table below

#ModelEffortScoreCost / taskHarness
1Qwen 2.5 7B–0s––
2Llama 3.2 1bnone0s––
3Llama 3.2 3bnone0s––
4Mistral Small 3none0s––
5Qwen 3.6 Flashmax0s––
6Mistral Medium 3.5high0s––
7Llama 3.1 70bnone0s––
8Qwen 2.5 72Bnone0s––
9Mistralnone0s––
10Mistral Medium 3none0s––
11Llama 3.1 8bmedium0s––
12Mistral Small 4high0s––
13Mistral Small 3.1none0s––
14Mistral Large 2407–0s––
15gpt-oss-120bhigh0s––
16Codestral 2508–1s––
17Gemini 2.5 Flash Litenone1s––
18Mistral Medium 3.1none1s––
19gpt-oss-20bhigh1s––
20Qwen 3 235B-A22Bxhigh1s––
21Mistral Small 3.2none1s––
22Inklingmax1s––
23Qwen 3.5 Flashmax1s––
24GPT-3.5high1s––
25Llama 4 Maverickmedium1s––
26Qwen 3 30B-A3Bnone1s––
27Llama 3.3 70bnone1s––
28Haiku 4.5xhigh1s––
29Mistral Largenone1s––
30Gemma 4 26Bxhigh1s––
31Gemma 2 27B–1s––
32Inkling Smallxhigh1s––
33Mistral Large 3none1s––
34Qwen 3 14Bnone1s––
35Qwen 2.5–1s––
36Gemini 2.5 Flashmedium1s––
37Qwen 3 Codermedium1s––
38Grok 4.20xhigh1s––
39Gemini 3.1 Flash-Litehigh1s––
40GPT-4.1 Nanohigh1s––
41GPT-4high1s––
42Gemma 3 4Bnone1s––
43DeepSeek V3.1none1s––
44GLM 4.7 Flashnone1s––
45DeepSeek V4.1 Flashmax1s––
46Gemma 4 31Bmax1s––
47Qwen 3.5 35B-A3Blow1s––
48Qwen 3high1s––
49Qwen 3.7 Flashmax1s––
50Llama 4 Scoutmedium1s––
51Qwen 3 32Bnone1s––
52Qwen 3.8 27Bmax1s––
53GPT-5.4 Nanoxhigh1s––
54Gemini 3.5 Flash-Litexhigh1s––
55Qwen 3.5 9Blow1s––
56Devstral 2none1s––
57Qwen 3.6 Plusmax1s––
58GLM 5.2max1s––
59GPT-4ohigh1s––
60Sonnet 4.6max1s––
61Qwen 3.5 Plusnone1s––
62GLM 5.3 Flashmax1s––
63GPT-4o minihigh1s––
64MiniMax M1–1s––
65Gemma 3 27Bnone1s––
66DeepSeek V4 Flash 0731max1s––
67Opus 4.7max1s––
68GPT-5.4 Minixhigh1s––
69Qwen 3.6 35B-A3Bmax1s––
70Qwen 3.6 27Bmax1s––
71GPT-4.1 Minihigh1s––
72Gemma 3 12Bnone1s––
73GLM 5.3max1s––
74MiniMax M2.1–1s––
75Mistral Large 4high1s––
76Qwen 3.7 Maxmax1s––
77Kimi K2.5max1s––
78Qwen 3.5 122B-A10Bnone1s––
79DeepSeek V4 Pro 0813max1s––
80GLM 4.5 Air–1s––
81GPT-5.1high1s––
82Opus 4.8max1s––
83Kimi K2medium1s––
84GPT-4.1high1s––
85MiniMax M2.5high1s––
86MiniMax M2.7high1s––
87Kimi K2.6max1s––
88Gemini 3.6 Flashxhigh1s––
89MiniMax M2–1s––
90MiniMax M3max1s––
91Gemini 3 Flashhigh1s––
92Qwen 3.5 397B-A17Bmax1s––
93DeepSeek V4 Flash 0423–1s––
94Qwen 3.7 Plushigh1s––
95GPT-5.4xhigh1s––
96Sonnet 4.5high1s––
97Qwen 3.8 Flashmax1s––
98Kimi K3max2s––
99DeepSeek V3none2s––
100Grok 4.3xhigh2s––
101Opus 4.5high2s––
102DeepSeek V4 Pro 0423–2s––
103Haiku 5.5max2s––
104Gemini 3.5 Flashxhigh2s––
105Sonnet 5.5max2s––
106Qwen 3.8xhigh2s––
107Gemini 3.1 Flash–2s––
108GLM 5high2s––
109Muse Spark 1.1max2s––
110Qwen 2.5 Codernone2s––
111Kimi K2.7 Codemax2s––
112GPT-5.2xhigh2s––
113Sonnet 5max2s––
114GLM 5.1max2s––
115Opus 4.6max2s––
116GPT-5.6 Lunamax2s––
117DeepSeek V4 Flashmax2s––
118Grok Build 0.1–2s––
119Qwen 3.5 27Bhigh2s––
120MiMo V2.5 Proxhigh2s––
121Gemini 3.7 Flashmax2s––
122Qwen 3.8 2.4T-A95B–2s––
123GPT-5 Nanohigh2s––
124DeepSeek V3.2high2s––
125Gemini 3.8 Flashmax2s––
126Grok 4.5max2s––
127GLM 4.6 Vlow2s––
128Hy3xhigh2s––
129Qwen 3.8 Maxmax2s––
130GPT-5.6 Terramax3s––
131GLM 4.5 Vnone3s––
132GPT-6 Lunamax3s––
133Gemini 3 Prohigh3s––
134o3high3s––
135GLM 4.7none3s––
136Opus 5max3s––
137Hy4–3s––
138GPT-5.2 Codexxhigh3s––
139Grok 4.7max3s––
140GPT-6 Solmax3s––
141GPT-5.6 Solmax3s––
142GLM 5 Vhigh3s––
143Muse Wondermax3s––
144Grok 4.6max3s––
145Opus 5.5max3s––
146MiMo V2.6 Flashhigh3s––
147o4 minihigh3s––
148o3 minihigh3s––
149Muse Spark 1.3max3s––
150Muse Spark 1.2max4s––
151Fable 5.1max4s––
152Gemini 3.1 Prohigh4s––
153o1high4s––
154MiMo V2.6 Prohigh4s––
155GPT-6.1 Solmax4s––
156Opus 4.1high4s––
157Fable 5max4s––
158Gemini 2.5 Prohigh4s––
159GLM 4.6none4s––
160GPT-5.3 Codexxhigh4s––
161MiMo V2.5xhigh4s––
162GPT-5 Minihigh4s––
163GPT-5.1 Codexmax5s––
164GPT-5.5xhigh5s––
165GPT-6 Astramax5s––
166Gemini 4max5s––
167GPT-5high5s––
168GPT-5.6 Terra Pro–6s––
169GPT-5.1 Codex Minihigh6s––
170GPT-5.6 Luna Pro–8s––
171GLM 4.5xhigh9s––
172GPT-6 Luna Promax12s––
173GPT-5.6 Sol Promax13s––
174GPT-5.5 Proxhigh14s––
175GPT-6 Astra Promax14s––
176GPT-6 Sol Promax14s––
177GPT-5.2 Proxhigh17s––
178o3 prohigh22s––
179GPT-6.1 Sol Pro–29s––