Benchmarks / Speed

Latency (Time to First Token)

Median seconds from request to first streamed chunk as measured by AA; may be a reasoning token. This is not time to first answer.

211models

–Open Index weight

–last updated

Artificial Analysispublisher

#	Model	Effort	Score	Cost / task	Harness
1	Gemini 2.5 Flash Lite	none	0s	–	–
2	North Mini Code	–	0s	–	–
3	Gemini 2.5 Flash	none	0s	–	–
4	Granite 4.2 3B	–	0s	–	–
5	Command A	–	1s	–	–
6	Haiku 4.5	none	1s	–	–
7	NVIDIA Nemotron 3 Nano 30B A3B (Non-reasoning)	none	1s	–	–
8	Qwen 3 30B-A3B	–	1s	–	–
9	NVIDIA Nemotron Nano 9B V2 (Reasoning)	–	1s	–	–
10	Nemotron 3.5 Lightning	–	1s	–	–
11	Celeris-1	none	1s	–	–
12	Grok 4.20	none	1s	–	–
13	GPT-5.4 Nano	none	1s	–	–
14	GPT-4.1 Nano	none	1s	–	–
15	Granite 4.2 8B	–	1s	–	–
16	Muse Spark Open (Xhigh)	xhigh	1s	–	–
17	Grok 4.3	none	1s	–	–
18	Qwen 3	–	1s	–	–
19	gpt-oss-20b	high	1s	–	–
20	Ministral 3 3B	none	1s	–	–
21	GPT-5.4 Mini	none	1s	–	–
22	DeepSeek V3	none	1s	–	–
23	GPT-4.1 Mini	none	1s	–	–
24	Qwen 3.5 9B	none	1s	–	–
25	GPT-5 Nano	minimal	1s	–	–
26	Nemotron 3 Nano Omni 30B A3B Reasoning	–	1s	–	–
27	GPT-5.6 Luna	none	1s	–	–
28	GPT-5.6 Terra	none	1s	–	–
29	GPT-5.4	none	1s	–	–
30	Gemini 3.5 Flash	minimal	1s	–	–
31	Ministral 3 8B	none	1s	–	–
32	GPT-5 Mini	minimal	1s	–	–
33	Gemma 4 E4B (Non-reasoning)	none	1s	–	–
34	GPT-5.5 Instant	–	1s	–	–
35	Phi-4 Mini Instruct	none	1s	–	–
36	Nova Micro	none	1s	–	–
37	Qwen 3 235B-A22B	–	1s	–	–
38	gpt-oss-120b	high	1s	–	–
39	Mistral Small 4	–	1s	–	–
40	GPT-5.2	none	1s	–	–
41	Nemotron 3.5 Super VL	–	1s	–	–
42	GPT-6 Sol	none	1s	–	–
43	Granite 4.2 30B	–	1s	–	–
44	GPT-5.5	none	1s	–	–
45	Sonnet 5	none	1s	–	–
46	Qwen 3.5 4B	–	1s	–	–
47	GPT-4.1	none	1s	–	–
48	Llama 3.1 8b	none	1s	–	–
49	Nova Lite	none	1s	–	–
50	DeepSeek V3.1	none	1s	–	–
51	Mistral Small 3.2	none	1s	–	–
52	Gemma 4 E4B (Reasoning)	–	1s	–	–
53	GPT-4o	none	1s	–	–
54	GPT-5.6 Sol	none	1s	–	–
55	GPT-4o mini	none	1s	–	–
56	Gemini 3 Flash	none	1s	–	–
57	GPT-6 Luna	none	1s	–	–
58	Qwen 3 14B	–	1s	–	–
59	NVIDIA Nemotron 3 Nano 30B A3B (Reasoning)	–	1s	–	–
60	Llama 4 Scout	none	1s	–	–
61	Mistral Small 3.1	none	1s	–	–
62	Gemma 4 26B	–	1s	–	–
63	Nova 2.0 Pro Preview (Non-reasoning)	none	1s	–	–
64	Mistral Small 3	none	1s	–	–
65	Qwen 3 Coder	none	1s	–	–
66	Ministral 3 14B	none	1s	–	–
67	Mistral Large 3	none	1s	–	–
68	Gemma 4 31B	–	1s	–	–
69	Sonnet 4.5	none	1s	–	–
70	NVIDIA Nemotron Nano 9B V2 (Non-reasoning)	none	1s	–	–
71	Nova Pro	none	1s	–	–
72	DeepSeek R1	–	1s	–	–
73	NVIDIA Nemotron Nano 12B v2 VL (Non-reasoning)	none	1s	–	–
74	Mixtral 8x7B Instruct	none	1s	–	–
75	Llama 4 Maverick	none	1s	–	–
76	Nova 2.0 Lite (Non-reasoning)	none	1s	–	–
77	GPT-5.1	none	1s	–	–
78	GPT-5	minimal	1s	–	–
79	DeepSeek V4 Flash	max	1s	–	–
80	Opus 4.5	none	1s	–	–
81	DeepSeek V4.1 Flash	max	1s	–	–
82	Gemini 3.7 Flash	low	1s	–	–
83	Sonnet 4.6	none	1s	–	–
84	Muse Glimmer (High)	high	1s	–	–
85	Kimi K2	none	1s	–	–
86	GLM 5.2	none	1s	–	–
87	Qwen 3 32B	none	1s	–	–
88	Trinity Large Thinking	–	1s	–	–
89	DeepSeek V4 Flash 0731	max	1s	–	–
90	MiniMax M3	–	1s	–	–
91	Sonnet 5.5	low	1s	–	–
92	Solar Pro 3	high	1s	–	–
93	Nemotron 3 Ultra	max	1s	–	–
94	Nemotron 3 Super 120B A12B (Reasoning)	–	1s	–	–
95	Quasar 438B (Max, Based on GLM-5.2)	max	1s	–	–
96	Opus 4.7	high	1s	–	–
97	MiniMax M2	–	1s	–	–
98	Agnes 2.5 Pro Beta	–	1s	–	–
99	MiniMax M2.5	–	1s	–	–
100	Reka Flash 3	–	1s	–	–
101	Llama 3.2 11b	none	2s	–	–
102	MiniMax M2.1	–	2s	–	–
103	Magistral Small 1.2	–	2s	–	–
104	GLM 5	none	2s	–	–
105	Mistral Large 4	–	2s	–	–
106	Solar Pro 4	–	2s	–	–
107	MiniMax M2.7	–	2s	–	–
108	GLM 4.7 Flash	none	2s	–	–
109	DeepSeek V4 Pro 0813	max	2s	–	–
110	DeepSeek V4 Pro 0424	none	2s	–	–
111	Grok Build 0.1	–	2s	–	–
112	GLM 5.1	–	2s	–	–
113	Opus 5	low	2s	–	–
114	Llama 3.3 70b	none	2s	–	–
115	Inkling Small	–	2s	–	–
116	MiniMax M1	–	2s	–	–
117	Ling 3.1 Flash	–	2s	–	–
118	Grok 3	none	2s	–	–
119	DeepSeek V4 Flash 0420	max	2s	–	–
120	GLM 4.7	none	2s	–	–
121	Ling 3.0 Flash Fin	–	2s	–	–
122	Qwen 3.5	none	2s	–	–
123	Solar Mini 4	–	2s	–	–
124	Kimi K2.5	–	2s	–	–
125	Inkling	xhigh	2s	–	–
126	Grok 3 Mini	high	2s	–	–
127	Llama 3.1 70b	none	2s	–	–
128	Qwen 3.6 Plus	–	2s	–	–
129	Qwen 3.6 35B-A3B	–	2s	–	–
130	Qwen 3.5 35B-A3B	–	2s	–	–
131	Qwen 3.8 Flash	–	2s	–	–
132	MiMo V2.5 Pro	none	2s	–	–
133	Qwen 3.5 397B-A17B	–	2s	–	–
134	Step 5 Preview	–	2s	–	–
135	Qwen 2.5 Turbo	none	2s	–	–
136	Qwen 3.7 Plus	–	2s	–	–
137	Mistral Medium 3.5	high	2s	–	–
138	Qwen 3.7 Max	max	2s	–	–
139	Llama 3.1 405b	none	2s	–	–
140	Qwen 3.5 122B-A10B	none	2s	–	–
141	Gemma 4 12B	none	2s	–	–
142	GLM 4.5 Air	–	2s	–	–
143	Ling-flash-2.0	none	2s	–	–
144	Mistral Medium 3	none	2s	–	–
145	Step 3.7 Flash	high	2s	–	–
146	Opus 4.6	high	2s	–	–
147	Phi-4	none	3s	–	–
148	LongCat 2.0	–	3s	–	–
149	GLM 5.3 Flash	max	3s	–	–
150	GLM 4.5 V	–	3s	–	–
151	Fable 5.1	low	3s	–	–
152	Qwen 3.8 Max	max	3s	–	–
153	Hy3	–	3s	–	–
154	GLM 5.3	max	3s	–	–
155	Kimi K2.6	none	3s	–	–
156	Qwen 3.8 2.4T-A95B	–	3s	–	–
157	Seed-OSS-36B-Instruct	–	3s	–	–
158	Ling 3.0 Flash	–	3s	–	–
159	GPT-6 Astra	low	3s	–	–
160	GPT-4	none	3s	–	–
161	GPT-6.1 Sol	low	3s	–	–
162	Mercury 2.5	–	3s	–	–
163	Kimi K2.7 Code	–	3s	–	–
164	Step 3.5 Flash 2603	–	3s	–	–
165	Step 3.5 Flash	–	3s	–	–
166	GLM 4.6	–	3s	–	–
167	Kimi K3	low	3s	–	–
168	GLM 4.6 V	–	4s	–	–
169	Qwen 3.6 27B	none	4s	–	–
170	Ring-2.6-1T	xhigh	4s	–	–
171	Grok 4.7	low	4s	–	–
172	Qwen 3.8 27B	low	4s	–	–
173	Qwen 2.5 72B	none	4s	–	–
174	MiMo V2.6 Pro	–	4s	–	–
175	Gemini 3.1 Flash-Lite	high	4s	–	–
176	MiMo V2.5	–	4s	–	–
177	MiMo V2.6 Flash	–	5s	–	–
178	Qwen 3.5 27B	none	6s	–	–
179	o3	–	6s	–	–
180	Grok 4.6	low	7s	–	–
181	Haiku 5.5	low	7s	–	–
182	o3 mini	–	7s	–	–
183	Gemini 3.5 Flash-Lite	high	7s	–	–
184	GPT-5.1 Codex	high	7s	–	–
185	Grok 4	–	7s	–	–
186	Mercury 2	high	7s	–	–
187	Opus 5.5	low	8s	–	–
188	Grok 4.5	high	8s	–	–
189	Nova 2.0 Pro Preview (Low)	low	8s	–	–
190	Granite 3.3 8B (Non-reasoning)	none	9s	–	–
191	GPT-5 Codex	high	10s	–	–
192	Nova 2.0 Lite (Low)	low	10s	–	–
193	GPT-5.1 Codex Mini	high	11s	–	–
194	Nova 2.0 Pro Preview (Medium)	medium	13s	–	–
195	Muse Spark 1.2	xhigh	13s	–	–
196	Gemini 3.6 Flash	high	15s	–	–
197	Nova 2.0 Lite (Medium)	medium	17s	–	–
198	Gemini 3.1 Pro	–	17s	–	–
199	Nova 2.0 Lite (High)	high	17s	–	–
200	o4 mini	high	21s	–	–
201	Gemini 2.5 Pro	–	21s	–	–
202	Muse Spark 1.1	xhigh	22s	–	–
203	Gemini 3.8 Flash	high	24s	–	–
204	Muse Spark 1.3	xhigh	24s	–	–
205	Ling 3.0 Tiny	–	32s	–	–
206	Opus 4.8	max	34s	–	–
207	Muse Wonder	max	42s	–	–
208	Gemini 4	high	45s	–	–
209	GPT-5.3 Codex	xhigh	70s	–	–
210	GPT-5.2 Codex	xhigh	75s	–	–
211	Fable 5	max	99s	–	–