Subject · tested 27 Oct 2026

Best AI models for economics

Scores on the 9 economics questions in the suite (4 from SuperGPQA, 5 from MMLU-Pro). That is a small sample: the bars show each score's 95% interval, and overlapping bars mean the models can't be told apart.

Questions9of 80 in the suite
Top score8/9Inkling, GPT-6 Astra, Muse Wonder, Grok 4.7, GPT-5.3-Codex, GPT-6.1 Terra Pro, Mercury 2.5, DeepSeek V4 Flash Vision Exp, Qwen3.8 27B, Gemini 3.1 Pro Preview, DeepSeek V4 Pro 0813, Muse Spark Open, Claude Opus 5.5, Gemini 3.8 Flash, GPT-6.1 Sol, Gemini 4 Argon, GPT-5.4, Claude Haiku 5.5, MiniMax M3.1 Flash, MiMo-V2.6-Pro, Kimi K3.1 Pro, Qwen3.8 Max (0902), Claude Sonnet 5.5
Tied with the leader69overlapping 95% intervals
Median score77.8%of 72 models

Economics: score by model

9 questions · the bar shows each score's 95% interval

Economics leaderboard
#ModelCorrectScore95% interval
1InklingThinking Machines8 / 988.9%56–98%
1GPT-6 AstraOpenAI · statistically tied with #18 / 988.9%56–98%
1Muse WonderMeta · statistically tied with #18 / 988.9%56–98%
1Grok 4.7SpaceXAI · statistically tied with #18 / 988.9%56–98%
1GPT-5.3-CodexOpenAI · statistically tied with #18 / 988.9%56–98%
1GPT-6.1 Terra ProOpenAI · statistically tied with #18 / 988.9%56–98%
1Mercury 2.5Inception · statistically tied with #18 / 988.9%56–98%
1DeepSeek V4 Flash Vision ExpDeepSeek · statistically tied with #18 / 988.9%56–98%
1Qwen3.8 27BQwen · statistically tied with #18 / 988.9%56–98%
1Gemini 3.1 Pro PreviewGoogle · statistically tied with #18 / 988.9%56–98%
1DeepSeek V4 Pro 0813DeepSeek · statistically tied with #18 / 988.9%56–98%
1Muse Spark OpenMeta · statistically tied with #18 / 988.9%56–98%
1Claude Opus 5.5Anthropic · statistically tied with #18 / 988.9%56–98%
1Gemini 3.8 FlashGoogle · statistically tied with #18 / 988.9%56–98%
1GPT-6.1 SolOpenAI · statistically tied with #18 / 988.9%56–98%
1Gemini 4 ArgonGoogle · statistically tied with #18 / 988.9%56–98%
1GPT-5.4OpenAI · statistically tied with #18 / 988.9%56–98%
1Claude Haiku 5.5Anthropic · statistically tied with #18 / 988.9%56–98%
1MiniMax M3.1 FlashMiniMax · statistically tied with #18 / 988.9%56–98%
1MiMo-V2.6-ProXiaomi · statistically tied with #18 / 988.9%56–98%
1Kimi K3.1 ProMoonshotAI · statistically tied with #18 / 988.9%56–98%
1Qwen3.8 Max (0902)Qwen · statistically tied with #18 / 988.9%56–98%
1Claude Sonnet 5.5Anthropic · statistically tied with #18 / 988.9%56–98%
24Step 5 PreviewStepFun · statistically tied with #17 / 977.8%45–94%
24Qwen3.8 2.4T A95BQwen · statistically tied with #17 / 977.8%45–94%
24Kimi K2.7 CodeMoonshotAI · statistically tied with #17 / 977.8%45–94%
24Qwen3.7 PlusQwen · statistically tied with #17 / 977.8%45–94%
24Claude Fable 5.1Anthropic · statistically tied with #17 / 977.8%45–94%
24o3OpenAI · statistically tied with #17 / 977.8%45–94%
24GPT-5.6 TerraOpenAI · statistically tied with #17 / 977.8%45–94%
24GPT-6.1 TerraOpenAI · statistically tied with #17 / 977.8%45–94%
24Solar Mini 4Upstage · statistically tied with #17 / 977.8%45–94%
24Ministral 3 8B 2512Mistral · statistically tied with #17 / 977.8%45–94%
24MiMo-V2.6-FlashXiaomi · statistically tied with #17 / 977.8%45–94%
24Hy3Tencent · statistically tied with #17 / 977.8%45–94%
24GLM 5.3 FlashZ.ai · statistically tied with #17 / 977.8%45–94%
24Qwen3 Coder NextQwen · statistically tied with #17 / 977.8%45–94%
24LongCat 2.0Meituan · statistically tied with #17 / 977.8%45–94%
24Inkling SmallThinking Machines · statistically tied with #17 / 977.8%45–94%
24Gemini 3.5 Flash LiteGoogle · statistically tied with #17 / 977.8%45–94%
24Mistral Large 4Mistral · statistically tied with #17 / 977.8%45–94%
24GLM 5.3Z.ai · statistically tied with #17 / 977.8%45–94%
24DeepSeek V4.1 Flash 1017DeepSeek · statistically tied with #17 / 977.8%45–94%
24GPT-6 SolOpenAI · statistically tied with #17 / 977.8%45–94%
24GPT-6 LunaOpenAI · statistically tied with #17 / 977.8%45–94%
24gpt-oss-120bOpenAI · statistically tied with #17 / 977.8%45–94%
24Llama 4 MaverickMeta · statistically tied with #17 / 977.8%45–94%
48Mistral Medium 3.5Mistral · statistically tied with #16 / 966.7%35–88%
48Command ACohere · statistically tied with #16 / 966.7%35–88%
48Nemotron 3.5 Super VLNVIDIA · statistically tied with #16 / 966.7%35–88%
48Ling 3.1 FlashinclusionAI · statistically tied with #16 / 966.7%35–88%
48gpt-oss-20bOpenAI · statistically tied with #16 / 966.7%35–88%
48Nova Micro 1.0Amazon · statistically tied with #16 / 966.7%35–88%
48Nemotron 3.5 LightningNVIDIA · statistically tied with #16 / 966.7%35–88%
48Mistral Small 4Mistral · statistically tied with #16 / 966.7%35–88%
48Trinity Large ThinkingArcee AI · statistically tied with #16 / 966.7%35–88%
48MiniMax M3MiniMax · statistically tied with #16 / 966.7%35–88%
48Nemotron 3 UltraNVIDIA · statistically tied with #16 / 966.7%35–88%
59Gemma 4 31BGoogle · statistically tied with #15 / 955.6%27–81%
59Qwen3.5 397B A17BQwen · statistically tied with #15 / 955.6%27–81%
59Hermes 4 405BNous · statistically tied with #15 / 955.6%27–81%
59Phi 4Microsoft · statistically tied with #15 / 955.6%27–81%
59Granite 4.2 8BIBM · statistically tied with #15 / 955.6%27–81%
59Llama 3.3 70B InstructMeta · statistically tied with #15 / 955.6%27–81%
59Solar Pro 4Upstage · statistically tied with #15 / 955.6%27–81%
59Qwen3.6 35B A3BQwen · statistically tied with #15 / 955.6%27–81%
67Llama 4 ScoutMeta · statistically tied with #14 / 944.4%19–73%
67Gemma 4 26B A4BGoogle · statistically tied with #14 / 944.4%19–73%
67Ministral 3 3B 2512Mistral · statistically tied with #14 / 944.4%19–73%
67Ministral 3 14B 2512Mistral · statistically tied with #14 / 944.4%19–73%
71Granite 4.0 MicroIBM2 / 922.2%6–55%
72Reka Flash 3rekaai0 / 90.0%0–30%

How hard are these questions?

Questions by the share of 72 scored models that got them right

  1. 1No model right0 questions
  2. 2Under 25% of models1 question
  3. 325–50% of models1 question
  4. 450–75% of models0 questions
  5. 575% or more, not all7 questions
  6. 6Every model right0 questions

Science →Engineering →Medicine →Law →Humanities & social science →

Economics: frequently asked questions

Which AI model is best at economics?
Inkling and GPT-6 Astra and Muse Wonder and Grok 4.7 and GPT-5.3-Codex and GPT-6.1 Terra Pro and Mercury 2.5 and DeepSeek V4 Flash Vision Exp and Qwen3.8 27B and Gemini 3.1 Pro Preview and DeepSeek V4 Pro 0813 and Muse Spark Open and Claude Opus 5.5 and Gemini 3.8 Flash and GPT-6.1 Sol and Gemini 4 Argon and GPT-5.4 and Claude Haiku 5.5 and MiniMax M3.1 Flash and MiMo-V2.6-Pro and Kimi K3.1 Pro and Qwen3.8 Max (0902) and Claude Sonnet 5.5 answered the most economics questions correctly: 8 of 9. With only 9 questions, 69 other models are statistically tied with the leader, so treat this ranking as a rough guide.
Where do the economics questions come from?
4 from SuperGPQA (ODC-BY 1.0) and 5 from MMLU-Pro (MIT). Every question is multiple choice with ten options and one correct answer.
How hard are the economics questions?
0 of the 9 questions were answered correctly by none of the 72 scored models, and 0 by all of them.