Skip to content
BetaBenchAlert v0.1 is in Beta.Numbers are real, but pages and rules can still change.See what changed

Live board

Models

The 4 models we lead with, ranked by typical total wait after at least 12 completed checks. Every other pin lives on Other models.

Typical total wait

The models we lead with

Each lab’s best model, plus Opus 5. Ranked by typical total wait on our machine; writing speed sits beside it.

  1. 1st place

    Grok 4.6Fastest

    grok-4.6 · xhigh

    6.44s

    total wait

    37.1 answer tok/s

    Sets the pace

  2. 2nd place

    Opus 5overloaded

    claude-opus-5 · high

    6.89s

    total wait

    35.0 answer tok/s

    7% longer wait

  3. 3rd place

    Fable 5

    claude-fable-5 · high

    7.79s

    total wait

    30.9 answer tok/s

    21% longer wait

  4. 4th place

    GPT-5.6 Sol

    gpt-5.6-sol · high

    10.9s

    total wait

    22.2 answer tok/s

    70% longer wait

Other modelsHaiku 4.5, Sonnet 5, GPT-5.6 Luna, GPT-5.6 Terra, GPT-5.5, and Grok 4.5. Same tests, every round. They are not on the home boards.6 pinned models

Read next

How to read this

An empty typical result means the model has fewer than 12 completed checks in the window. It never means zero, and it never means broken.