Benchmarks / Artificial Analysis

Reported by Artificial Analysis

Artificial Analysis

Median seconds from sending a request to the first token of the answer, including any thinking, measured by Artificial Analysis on the model's usual API.

Last updated 20 Oct 2026

Results dated
20 Oct 2026
Results
137 configurations of 78 models
Unit
seconds
Licence
Artificial Analysis commercial data licence

Time to first answer token: Beam 1.0 501B A23B

Top 15 of 78 results · seconds, lower is better · ≈ cannot be told apart from the leader. Choose a model to highlight it.Clear highlight

  1. 1≈ Nemotron 3 Nano 30B A3B (no reasoning)NVIDIA 0.36
  2. 1≈ Command ACohere 0.36
  3. 3 Ministral 3 3BMistral AI 0.41
  4. 4 Qwen3.5-9B (no reasoning)Alibaba 0.43
  5. 4 Gemma 4 E4B (no reasoning)Google 0.43
  6. 6 Ministral 3 8BMistral AI 0.46
  7. 7 Ministral 3 14BMistral AI 0.52
  8. 8 Nova Micro 1.0Amazon 0.55
  9. 9 Mistral Small 4 (no reasoning)Mistral AI 0.57
  10. 9 Qwen3-Next-80B-A3B-InstructAlibaba 0.57
  11. 11 Gemma 4 26B A4B (no reasoning)Google 0.60
  12. 12 Llama 4 ScoutMeta 0.61
  13. 13 DeepSeek-V4.1-Flash (no reasoning)DeepSeek 0.63
  14. 14 Llama 3.3 70B InstructMeta 0.64
  15. 15 Llama 4 MaverickMeta 0.65
  16. 42 Beam 1.0 501B A23BReflection 6.73

Full results

Artificial Analysis: Time to first answer token, seconds, lower is better
#ModelTime to first answer token
seconds, lower is better		Price
$ per million tokens, in / out
1≈ Nemotron 3 Nano 30B A3B (no reasoning)NVIDIA · best of 2 settings
0.36
$0.05 / $0.20
1≈ Command ACohere
0.36
$2.50 / $10
3 Ministral 3 3BMistral AI
0.41
$0.10 / $0.10
4 Qwen3.5-9B (no reasoning)Alibaba · best of 2 settings
0.43
$0.10 / $0.15
4 Gemma 4 E4B (no reasoning)Google · best of 2 settings
0.43
–
6 Ministral 3 8BMistral AI
0.46
$0.15 / $0.15
7 Ministral 3 14BMistral AI
0.52
$0.20 / $0.20
8 Nova Micro 1.0Amazon
0.55
$0.035 / $0.14
9 Mistral Small 4 (no reasoning)Mistral AI · best of 2 settings
0.57
$0.15 / $0.60
9 Qwen3-Next-80B-A3B-InstructAlibaba
0.57
$0.10 / $1.10
11 Gemma 4 26B A4B (no reasoning)Google · best of 2 settings
0.60
$0.10 / $0.30
12 Llama 4 ScoutMeta
0.61
$0.18 / $0.59
13 DeepSeek-V4.1-Flash (no reasoning)DeepSeek · best of 2 settings
0.63
$0.30 / $1.20
14 Llama 3.3 70B InstructMeta
0.64
$0.59 / $0.79
15 Llama 4 MaverickMeta
0.65
$0.27 / $0.85
16 Qwen3.5-4B (no reasoning)Alibaba · best of 2 settings
0.70
–
17 Claude Sonnet 5.5 (low reasoning)Anthropic · best of 5 settings
0.71
$2 / $10
18 GPT-5.6 Terra (no reasoning)OpenAI · best of 6 settings
0.72
$2 / $12
19 Nova 2 Lite (no reasoning)Amazon · best of 4 settings
0.73
$0.30 / $2.50
20 GPT-6 Luna (no reasoning)OpenAI · best of 5 settings
0.74
$0.10 / $0.50
21 Qwen3-Coder-NextAlibaba
0.75
$0.18 / $0.90
22 Gemma 4 31B (no reasoning)Google · best of 2 settings
0.77
$0.14 / $0.40
23 Qwen3-Omni-30B-A3B-InstructAlibaba
0.87
–
24 Claude Fable 5.1 (low reasoning)Anthropic · best of 5 settings
0.92
$10 / $50
25 Qwen3.5-Omni-FlashAlibaba
0.94
–
26 Qwen3.5-122B-A10B (no reasoning)Alibaba · best of 2 settings
0.99
$0.26 / $2.08
27 Phi-4Microsoft
1.03
$0.07 / $0.14
28 Qwen3.6-35B-A3B (no reasoning)Alibaba · best of 2 settings
1.09
$0.10 / $1
29 Gemma 4 12B (no reasoning)Google · best of 2 settings
1.15
–
29 Qwen3.5-Omni-PlusAlibaba
1.15
–
31 DeepSeek-V4-Pro (0813, no reasoning)DeepSeek · best of 2 settings
1.18
$1.32 / $3.96
32 Qwen3.8-27B (no reasoning)Alibaba · best of 4 settings
1.27
$0.50 / $3
33 GPT-6 Astra (low reasoning)OpenAI · best of 5 settings
1.41
$10 / $50
34 Qwen3.5-397B-A17B (no reasoning)Alibaba · best of 2 settings
1.54
$0.55 / $3.50
35 GPT-6.1 Sol (low reasoning)OpenAI · best of 5 settings
1.71
$2 / $10
36 Grok 4.7 (low reasoning)xAI · best of 3 settings
4.00
$2 / $6
37 o3OpenAI
4.60
$2 / $8
38 Gemini 3.5 Flash-LiteGoogle
4.69
$0.30 / $2.50
39 Nemotron 3.5 Super VLNVIDIA
5.43
–
40 Claude Haiku 5.5 (low reasoning)Anthropic · best of 5 settings
6.41
$0.10 / $0.50
41 Trinity Large ThinkingArcee AI
6.63
$0.25 / $0.80
42 Beam 1.0 501B A23BReflection
6.73
–
43 Claude Opus 5.5 (low reasoning)Anthropic · best of 5 settings
6.89
$4 / $20
44 gpt-oss-20b (low reasoning)OpenAI · best of 2 settings
7.62
$0.03 / $0.15
45 gpt-oss-120b (high reasoning)OpenAI · best of 2 settings
8.22
$0.15 / $0.60
46 DeepSeek-V4.1-Flash-1017 (max reasoning)DeepSeek
10.5
$0.30 / $1.20
47 Inkling (extra-high reasoning)Thinking Machines
10.9
$0.95 / $4.05
48 DeepSeek-V4-Flash-Vision-Exp (max reasoning)DeepSeek
11.6
$0.44 / $1.32
49 Qwen3-Next-80B-A3B-ThinkingAlibaba
12.7
$0.15 / $1.20
50 Inkling SmallThinking Machines
12.9
$0.45 / $1.20
51 MiniMax M3.1 FlashMiniMax
13.3
$0.15 / $0.60
51 Mistral Medium 3.5Mistral AI
13.3
$1.50 / $7.50
53 GPT-5.5 Instant (2026-06-26)OpenAI
13.9
–
54 Muse Glimmer (high reasoning)Meta
14.5
–
55 MiniMax-M3MiniMax
14.6
$0.30 / $1.20
55 Muse Spark Open (extra-high reasoning)Meta
14.6
–
57 Gemini 3.1 Pro PreviewGoogle
14.7
$2 / $12
58 Gemini 3.8 Flash (high reasoning)Google
15.5
$1.50 / $7.50
59 Solar Pro 4Upstage
17.9
$0.09 / $0.36
60 Qwen3-Omni-30B-A3B-ThinkingAlibaba
19.4
–
61 Reka Flash 3Reka AI
24.8
$0.10 / $0.20
62 GLM-5.3 (low reasoning)Z.ai · best of 2 settings
25.4
$1.40 / $4.40
63 Hy3Tencent
25.8
$0.14 / $0.58
64 Magistral Small 1.2Mistral AI
27.1
–
65 GLM-5.3-FlashZ.ai
27.5
$0.15 / $0.50
66 Muse Spark 1.3 (extra-high reasoning)Meta · best of 2 settings
29.6
$1.25 / $4.25
67 Granite 4.2 8BIBM
31.2
$0.06 / $0.25
68 Mistral Large 4Mistral AI
33.9
$1.36 / $4.18
69 Qwen3.7-PlusAlibaba
37.3
$0.32 / $1.28
70 Qwen3.8-Flash-NextAlibaba
37.6
–
71 Kimi K2.7 CodeMoonshot AI
42.7
$0.95 / $4
72 GPT-5.3-Codex (extra-high reasoning)OpenAI
44.9
$1.75 / $14
73 MiMo-V2.6-FlashXiaomi
47.4
$0.14 / $0.28
74 Muse Wonder (max reasoning)Meta
48.2
$4 / $20
75 MiMo-V2.6-ProXiaomi
49.8
$0.43 / $0.87
76 Kimi K3 (max reasoning)Moonshot AI · best of 2 settings
51.8
$3 / $15
77 Qwen3.8-Max (0902)Alibaba
56.2
$2 / $6
78 Qwen3.8-2.4T-A95BAlibaba
58.0
$2 / $6

Swipe the table sideways for more columns.

Ranks follow the score as shown, so equal numbers share a rank. Each model is shown at its best setting; show every setting. Results as published by Artificial Analysis; we do not re-run them.

Progress over time

Time to first answer token, plotted at each model's release date, lower is better. Each model counts once, at its best setting. Scores are today's, not launch day's. Hover or tap a mark for its model.

Best score so far

Every model at its release date. The line moves each time a new release beats every earlier one, and each new best is named where there is room.

-20020406080 2025Jul2026Jul Nova Micro 1.0Ministral 3 3BNemotron 3 Nano 30B A3B

Best score by lab

Each lab's best so far, stepping whenever one of its releases beats its own earlier models. The 6 labs with the best scores today; 17 others not drawn.

NVIDIACohereMistral AIAlibabaGoogleAmazon

-100102030 2025Jul2026Jul NVIDIACohereMistral AIAlibabaGoogleAmazon

What it measures

Median seconds from sending a request to the first token of the answer, including any thinking, measured by Artificial Analysis on the model's usual API.

What it does not measure

Not total time for a long answer, or latency on other hosts.

Source

Artificial Analysis · Methodology

285 results from Artificial Analysis not ranked here · show why

We rank a result only when we can tie it to a specific model you can use. These are left out:

  • Not on sale through the API providers we track: 270
  • A different snapshot or variant from the model we list: 14
  • An unusual combination of settings: 1

Data sourced from Artificial Analysis. Licence: Artificial Analysis commercial data licence.