Skip to main content
Meo

Median time to first token (s) — AI model leaderboard

AI models ranked by Median time to first token (s), an aggregated third-party benchmark from artificial_analysis. Lower is better. Cross-referenced against our first-party meo scores and Effective Value (𝕍).

Ranking 87 models across the full field · as of 2026-06-07.

#ModelLabMedian time to first token (s)
1Z.ai: GLM 5 Turboz-ai0
2Z.ai: GLM 5V Turboz-ai0
3inclusionAI: Ling-2.6-1Tinclusionai0
4Google: Gemma 4 26B A4B (free)google0
5inclusionAI: Ling-2.6-flashinclusionai0
6Upstage: Solar Pro 3upstage0
7OpenAI: o1-proopenai0
8OpenAI: o1openai0
9Prime Intellect: INTELLECT-3prime-intellect0
10Google: Gemini 2.5 Flash Lite Preview 09-2025google0
11Google: Gemma 3 27Bgoogle0
12Google: Gemma 3 12Bgoogle0
13OpenAI: GPT-3.5 Turbo (older v0613)openai0
14OpenAI: GPT-5.4 Proopenai0
15OpenAI: GPT-5.5 Proopenai0
16Cohere: Command Acohere0.31
17OpenAI: GPT-4.1 Nanoopenai0.39
18OpenAI: gpt-oss-20bopenai0.42
19IBM: Granite 4.1 8Bibm-granite0.42
20OpenAI: GPT-4oopenai0.49
21Google: Gemini 2.5 Flashgoogle0.5
22Microsoft: Phi 4microsoft0.5
23OpenAI: gpt-oss-120bopenai0.53
24OpenAI: GPT-4.1 Miniopenai0.57
25OpenAI: GPT-4o (2024-05-13)openai0.57
26OpenAI: GPT-4o-miniopenai0.57
27OpenAI: GPT-4.1openai0.61
28Meta: Llama 4 Scoutmeta-llama0.62
29Meta: Llama 4 Maverickmeta-llama0.63
30OpenAI: GPT-4o (2024-08-06)openai0.65
31Mistral: Mistral Medium 3.5mistralai0.65
32Arcee AI: Trinity Large Thinkingarcee-ai0.77
33Z.ai: GLM 5.1z-ai0.86
34Qwen: Qwen3.5-9Bqwen0.93
35DeepSeek: DeepSeek V4 Flashdeepseek0.95
36Google: Gemini 3 Flash Previewgoogle0.95
37Google: Gemma 4 31Bgoogle1.02
38OpenAI: GPT-4openai1.05
39Kwaipilot: KAT-Coder-Pro V2kwaipilot1.07
40Anthropic: Claude Sonnet 4.6anthropic1.08
41DeepSeek: DeepSeek V4 Prodeepseek1.19
42Qwen: Qwen3.5-122B-A10Bqwen1.23
43MiniMax: MiniMax M2.7minimax1.23
44Qwen: Qwen3.6 35B A3Bqwen1.24
45Qwen: Qwen3 Coder Nextqwen1.31
46Qwen: Qwen3.7 Plusqwen1.32
47StepFun: Step 3.7 Flashstepfun1.32
48Xiaomi: MiMo-V2-Flashxiaomi1.37
49Microsoft: Phi 4 Mini Instructmicrosoft1.39
50Qwen: Qwen3.6 27Bqwen1.45
51StepFun: Step 3.5 Flashstepfun1.5
52OpenAI: GPT-4 Turboopenai1.68
53Qwen: Qwen3.7 Maxqwen1.72
54Qwen: Qwen3.6 Plusqwen1.73
55Qwen: Qwen3.5 397B A17Bqwen1.9
56inclusionAI: Ring-2.6-1Tinclusionai1.91
57MiniMax: MiniMax M3minimax2.38
58Xiaomi: MiMo-V2.5-Proxiaomi2.39
59Tencent: Hy3 previewtencent2.53
60OpenAI: GPT-5.2-Codexopenai2.54
61Inception: Mercury 2inception3.05
62OpenAI: GPT-5.4 Nanoopenai3.31
63OpenAI: GPT-5.1-Codex-Miniopenai4.09
64OpenAI: GPT-5 Codexopenai4.59
65OpenAI: GPT-5.1-Codexopenai5.19
66Google: Gemini 3.1 Flash Litegoogle5.26
67OpenAI: o3 Miniopenai5.71
68OpenAI: o3openai5.81
69OpenAI: GPT-5.4 Miniopenai6.74
70Reka Flash 3rekaai15
71Google: Gemini 3.5 Flashgoogle16
72xAI: Grok 4.3x-ai17
73Google: Gemini 2.5 Progoogle19
74OpenAI: o4 Miniopenai20
75OpenAI: o3 Mini Highopenai20
76OpenAI: GPT-5.1openai21
77Anthropic: Claude Opus 4.7anthropic24
78Google: Gemini 3.1 Pro Previewgoogle25
79Anthropic: Claude Opus 4.8anthropic39
80OpenAI: GPT-5.3-Codexopenai59
81OpenAI: GPT-5.5openai63
82OpenAI: o3 Proopenai64
83OpenAI: GPT-5 Miniopenai83
84OpenAI: GPT-5openai83
85OpenAI: GPT-5 Nanoopenai88
86OpenAI: GPT-5.2 Chatopenai115
87OpenAI: GPT-5.4openai178

Artificial Analysis (artificialanalysis.ai). Redistribution requires an AA commercial license.

← All rankingsMethodology & 𝕍 →