AI model rankings

Live LLM rankings based on benchmarks and real routing data from teams using models through SwitchGate. All figures below are placeholder data until the live catalog ships.

Top models

Weekly usage across SwitchGate

Tokens routed per week, stacked by model, over the last twelve months.

    LLM leaderboard

    The most-routed models this week

    Ranked by tokens routed through the gateway over the last seven days.

    Browse all models
    1

    deepseek-v3.2

    by DeepSeek

    10.0T tokens

    ↑191%

    2

    gpt-5.2

    by OpenAI

    9.48T tokens

    ↑90%

    3

    claude-sonnet-4-6

    by Anthropic

    5.5T tokens

    ↓21%

    4

    gemini-3-flash

    by Google

    5.03T tokens

    ↓1%

    5

    gpt-5-mini

    by OpenAI

    4.88T tokens

    ↑63%

    6

    llama-4-scout

    by Groq

    3.69T tokens

    ↑28%

    7

    mistral-large-3

    by Mistral

    2.55T tokens

    ↓14%

    8

    grok-4-fast

    by xAI

    2.46T tokens

    ↑451%

    9

    nova-premier

    by AWS Bedrock

    2.24T tokens

    ↓4%

    10

    qwen3-235b

    by Alibaba

    1.76T tokens

    ↑23%

    Placeholder rankings for design purposes; live telemetry ships with the API.

    Route to any of them with one line

    Every model on this page is one model= string away, behind your budgets and locks.