❤️❤️❤️❤️❤️❤️ 我们已经正式推出微信小程序，在微信中搜索 TrendForge Pro 即可使用小程序，如果使用 Telegram 请搜索 trendforge_tg ❤️❤️❤️❤️❤️❤️

AI 模型排行

实时追踪主流大语言模型的使用量、性能与基准测试数据

模型使用排行 基准测试 性能指标 应用排行 AI 厂商动态
按吞吐排序 按延迟排序
# 模型 P50 延迟 P50 吞吐
1 openai/gpt-oss-120b
openai
 150 ms 782.5 t/s
2 openai/gpt-oss-safeguard-20b
openai
 189 ms 665.0 t/s
3 openai/gpt-oss-20b
openai
 157 ms 509.5 t/s
4 qwen/qwen3-32b-04-28
qwen
 311 ms 481.0 t/s
5 inception/mercury-2-20260304
inception
 479 ms 439.0 t/s
6 z-ai/glm-4.7-20251222
z-ai
 356 ms 332.0 t/s
7 minimax/minimax-m2.5-20260211
minimax
 331 ms 302.0 t/s
8 meta-llama/llama-3.1-8b-instruct
meta-llama
 71 ms 270.0 t/s
9 meta-llama/llama-4-scout-17b-16e-instruct
meta-llama
 297 ms 165.0 t/s
10 google/gemini-2.5-flash-image
google
 827 ms 164.5 t/s
11 openai/o3-mini-2025-01-31
openai
 4386 ms 156.0 t/s
12 nvidia/nemotron-3-nano-30b-a3b
nvidia
 393 ms 153.0 t/s
13 amazon/nova-micro-v1
amazon
 422 ms 142.0 t/s
14 qwen/qwen3.6-flash
qwen
 836 ms 138.0 t/s
15 meta-llama/llama-3.3-70b-instruct
meta-llama
 190 ms 138.0 t/s
16 qwen/qwen3.6-35b-a3b-20260415
qwen
 456 ms 134.0 t/s
17 moonshotai/kimi-k2.6-20260420
moonshotai
 568 ms 134.0 t/s
18 google/gemini-2.5-flash-lite
google
 448 ms 124.0 t/s
19 qwen/qwen3-next-80b-a3b-instruct-2509
qwen
 450 ms 120.0 t/s
20 mistralai/devstral-small-2507
mistralai
 279 ms 113.0 t/s
21 qwen/qwen3-coder-next-2025-02-03
qwen
 423 ms 112.0 t/s
22 google/gemini-3.1-flash-lite-preview-20260303
google
 708 ms 110.0 t/s
23 amazon/nova-2-lite-v1
amazon
 1196 ms 104.0 t/s
24 openai/gpt-5-nano-2025-08-07
openai
 3205 ms 104.0 t/s
25 google/gemini-3.1-flash-image-preview-20260226
google
 12192 ms 104.0 t/s
26 moonshotai/kimi-k2-0905
moonshotai
 252 ms 102.0 t/s
27 qwen/qwen3.5-9b-20260310
qwen
 411 ms 100.0 t/s
28 qwen/qwen3.5-35b-a3b-20260224
qwen
 363 ms 100.0 t/s
29 google/gemini-3.1-flash-lite-20260507
google
 785 ms 98.0 t/s
30 anthropic/claude-4.5-haiku-20251001
anthropic
 461 ms 98.0 t/s
31 mistralai/mistral-small-2603
mistralai
 485 ms 98.0 t/s
32 poolside/laguna-xs.2-20260421
poolside
 631 ms 97.0 t/s
33 stepfun/step-3.5-flash
stepfun
 494 ms 97.0 t/s
34 openai/gpt-5-chat-2025-08-07
openai
 735 ms 97.0 t/s
35 minimax/minimax-m2.7-20260318
minimax
 1037 ms 90.0 t/s
36 google/gemini-2.0-flash-lite-001
google
 388 ms 89.0 t/s
37 google/gemini-2.5-pro
google
 2063 ms 87.0 t/s
38 z-ai/glm-5-20260211
z-ai
 628 ms 87.0 t/s
39 qwen/qwen3.5-flash-20260224
qwen
 526 ms 86.0 t/s
40 qwen/qwen3-8b-04-28
qwen
 343 ms 86.0 t/s
41 sao10k/l3-lunaris-8b
sao10k
 139 ms 85.0 t/s
42 openai/gpt-5-mini-2025-08-07
openai
 3724 ms 85.0 t/s
43 arcee-ai/trinity-large-thinking
arcee-ai
 554 ms 85.0 t/s
44 deepseek/deepseek-v4-flash-20260423
deepseek
 552 ms 84.0 t/s
45 google/gemini-2.5-flash-lite-preview-09-2025
google
 1050 ms 80.0 t/s
46 qwen/qwen-2.5-7b-instruct
qwen
 356 ms 77.0 t/s
47 mistralai/mistral-nemo
mistralai
 296 ms 77.0 t/s
48 qwen/qwen3.5-397b-a17b-20260216
qwen
 618 ms 77.0 t/s
49 mistralai/mistral-small-3.2-24b-instruct-2506
mistralai
 325 ms 77.0 t/s
50 google/gemini-3-pro-image-preview-20251120
google
 3501 ms 77.0 t/s
助手