AI Coding Models

OpenAI

• GPT-5.6-Luna

  Not tested

• GPT-5.6-Sol

  Not tested

• GPT-5.6-Terra

  Not tested

• GPT-6-Astra

  #4

• GPT-6-Luna

  #14

• GPT-6-Sol

  #13

• GPT-6.1-Sol

  #1

Anthropic

• Fable 5.1

  #9

• Haiku 5.5

  #15

• Opus 5

  Not tested

• Opus 5.5

  #2

• Sonnet 5

  Not tested

• Sonnet 5.5

  #8

• Opus 4.8

  Not tested

xAI

• Grok 4.5

  Not tested

• Grok 4.6

  #20

• Grok 4.7

  #12

Stealth

• Omen Alpha

  Not tested

• Ox Alpha

  Not tested

  This model was a stealth codename for the model GLM-5.3-Flash, released on August 26, 2026.

• Space Bunny

  #17

Devin

• SWE-2

  #18

Moonshot AI

• Kimi K3

  #19

• Kimi K2.7 Code

  Not tested

Z.ai

• GLM-5.3

  #23

• GLM-5.3-Flash

  #36

• GLM-5.2

  Not tested

DeepSeek

• Deepseek-V4-Pro-0813

  #32

• Deepseek-V4.1-Flash

  #24

• Deepseek-V4-Flash

  Not tested

Xiaomi

• MiMo-V2.6 Flash

  #28

• MiMo-V2.6 Pro

  #25

Alibaba

• Qwen 3.8 27B

  #35

• Qwen 3.8 Flash

  #31

• Qwen 3.8 Max

  Not tested

• Qwen 3.8 Max (0902)

  #26

Meta

• Muse Spark 1.2

  Not tested

• Muse Spark 1.3

  #30

Tencent

• Tencent Hy3

  #33

• Tencent Hy4 Preview

  Not tested

  This is a preview model by Tencent, announced on August 28, 2026. I have not run the full LLM coding benchmark with it, because, after just 1 prompt out of 20,...

Google

• Gemini-3.6-Flash

  Not tested

• Gemini-3.7-Flash

  Not tested

• Gemini-3.8-Flash

  #37

• Gemini-3.1-Pro

  #40

Longcat

• LongCat-2.5-Preview

  #38

Cursor

• Composer 2.5

  #41

MiniMax

• Minimax M3

  #42