Leaderboard

Personal AI Tier List

My Current AI Rankings

Based on my own tests, my own workflow, and the models I can actually run or use right now.

Codex Harness / Cloud Coding

Agent
  1. 1
    ChatGPT 5.5 Codex Harness My #1 coding-agent result.
  2. 2
    Claude Opus 4.7 Strong long-task coding and repo reasoning.
  3. 3
    GLM-5.1 Strong modern coding and agent model.

Local Coding

Hardware Limited
  1. 1
    Qwen3.5-9B Q8 My current best local coding model.
  2. 2
    Gemma 4 E4B Small reasoning/coding model worth testing locally.
  3. 3
    Qwopus Dual Reasoning A finetuned coding model
Local disclaimer: I cannot properly test larger local models right now, so this ranking is limited to my current hardware.

Image Models

Visual
  1. 1
    GPT Image 2 My current pick for best image model.
  2. 2
    Nano Banana Pro My second-place image model.
  3. 3
    Z-Image Turbo Fast and useful image model.
  4. 4
    FLUX.2 [klein] 9B Good image model, but less reliable with bodies in my tests.
FLUX.2 [klein] 9B caveat: in my testing, it has horrible body deformations around 30% of the time.
Note: This is my personal ranking, not a universal benchmark. Cloud rankings are based on my coding-agent results. Local rankings are based on what I can realistically run.