ai benchmarks

Real-world AI performance metrics

Gauntlet Composite updated

Gauntlet Science updated

Gauntlet Ship updated

Gauntlet Teach updated

Gauntlet Judgment updated

Artificial Analysis data updated

Every change Artificial Analysis made to its scores and prices, newest first. Click a model to see its past scores.

1 new release · 2 removed releases · 11 price changes

Inkling (25) and MiniMax-M2.7 (22.8) were removed, while Trinity Large Thinking entered at 10.8 intelligence and $0.12 per task. Costs rose for Mercury 2.5 ($0.0642 to $0.1167), GLM-5.2 max ($0.9649 to $1.473), Claude 4.5 Haiku ($0.2077 to $0.2768), and several others. Costs fell for Mistral Large 3 ($0.0538 to $0.0311), Qwen3.7 Plus ($0.2952 to $0.217), and DeepSeek V4 Flash 0420 max ($0.1179 to $0.1075); Mistral’s two models moved in opposite directions.

AI-written summary of these changes · gpt-5.6-terra

New releases 1

  • peak 10.8one setting, $0.12 per task

Price changes 11

  • single settingCost per task$0.064 → $0.12+$0.053
  • single settingCost per task$0.054 → $0.031−$0.023
  • maxCost per task$0.96 → $1.47+$0.51
  • single settingCost per task$0.30 → $0.22−$0.078
  • single settingCost per task$0.21 → $0.28+$0.069
  • highCost per task$0.17 → $0.21+$0.040
  • single settingCost per task$0.44 → $0.50+$0.065
  • maxCost per task$0.12 → $0.11−$0.010
  • xhighCost per task$0.41 → $0.45+$0.038
  • single settingCost per task$0.55 → $0.60+$0.048
  • single settingCost per task$0.098 → $0.10+$0.0021

Removed releases 2

  • was 25.0 peakhad one setting
  • was 22.8 peakhad one setting

1 price change

StepFun's Step 5 Preview cost per Intelligence Index task increased from $0.7155 to $0.7175, a rise of $0.0020.

AI-written summary of these changes · gpt-5.6-terra

Price changes 1

  • single settingCost per task$0.72 → $0.72+$0.0020

1 price change

Z AI's GLM-5.3 low-effort cost per Intelligence Index task fell from $3.663 to $0.8522, a reduction of about 77%. No score, release, or reasoning-effort availability changes are listed.

AI-written summary of these changes · gpt-5.6-terra

Price changes 1

  • lowCost per task$3.66 → $0.85−$2.81

1 new release

Apodex 1.1 was added with an Intelligence Index score of 26.4 in its one listed setting. Its cost is $0.46 per Intelligence Index task.

AI-written summary of these changes · gpt-5.6-terra

New releases 1

  • peak 26.4one setting, $0.46 per task

1 price change · 1 new effort level

Z AI's GLM-5.3 added a low reasoning-effort result of 34.3 on the Intelligence Index at $3.66 per task. Meta's Muse Glimmer high-effort cost per Intelligence task rose from $0.0551 to $0.0567.

AI-written summary of these changes · gpt-5.6-terra

Price changes 1

  • highCost per task$0.055 → $0.057+$0.0016

New effort levels 1

  • lowIntelligence34.3$3.66 per task