GPT Load (AllAI)

GPT Load is an OpenAI-compatible API load balancing service, distributing requests across multiple AI model providers.

Categories

GPT Load (AllAI) offers 17 LLM API models.

Speed benchmark average: 269 tok/s.

GPT Load (AllAI) is an API aggregator, offering models from multiple vendors.

Avg Speed269.02 tok/s

Latency3.04 s

Total Tests153

Models17

Updated4/16/2026

Created At12/7/2025

Supported Models

Model	Speed	Latency	Tests
qwen-3-32b	2105.17 tok/s	1.93s	5
qwen-3-235b-a22b-instruct-2507	558.49 tok/s	1.49s	20
models/gemini-2.5-flash-lite	259.93 tok/s	0.71s	5
DeepSeek-V3.1	257.63 tok/s	0.55s	5
DeepSeek-V3-0324	218.28 tok/s	0.90s	5
openai/gpt-oss-120b	205.99 tok/s	8.90s	5
models/gemini-2.5-flash	198.97 tok/s	7.95s	15
models/gemini-2.5-flash-preview-09-2025	175.68 tok/s	0.57s	5
Qwen/Qwen3-Next-80B-A3B-Instruct	173.09 tok/s	1.54s	5
qwen/qwen3-next-80b-a3b-instruct	154.28 tok/s	0.79s	10
gpt-oss:120b	141.34 tok/s	1.35s	5
translate-model	138.91 tok/s	3.10s	45
x-ai/grok-4.1-fast:free	87.31 tok/s	6.31s	5
deepseek-v3.1:671b	72.01 tok/s	2.63s	5
WiNGPT-Babel	70.20 tok/s	1.85s	3
gpt-4.1-mini	66.95 tok/s	1.58s	5
moonshotai/kimi-k2-instruct-0905	36.98 tok/s	5.74s	5