AI Model Rankings

Real-world LLM usage on 3srouter. Models are ranked by tokens processed through the 3srouter API.

3srouter usage data · Through October 19, 2026. View source

Top models

Total tokens on 3srouter

DeepSeek V4.1 Flash
29.6T
GLM 5.3 Flash
11.9T
MiMo-V2.6-Flash
9.57T
Hy4 preview
7.82T
DeepSeek V4.1 Flash 1017
7.34T

LLM Leaderboard

Top models by task

Market Share

Share of all reference tokens for listed creators · Week of October 12

DeepSeek24.4%Z.ai12.2%Qwen5.3%

Fastest models

Observed throughput on 3srouter · October 20, 2026

ModelThroughputLatencyProviders
Qwen3.8 27B427 tok/s300 ms21
GLM 5.2402 tok/s327 ms34
DeepSeek V4.1 Flash346 tok/s248 ms36

Cost per session

Compare the cost of real-world model sessions.

View live data

This breakdown is available in the live 3srouter rankings.

Open / Closed

Track the share of open-weight and proprietary model usage.

View live data

This breakdown is available in the live 3srouter rankings.

Languages

Explore model usage across natural languages.

View live data

This breakdown is available in the live 3srouter rankings.

Programming

Coding-capable models from the public catalog

Explore collection

Domain

Explore how models are used across task domains.

View live data

This breakdown is available in the live 3srouter rankings.

Context Length

Tool Calls

Explore models

Public rankings data: Original source · CC BY 4.0 · October 20, 2026.