ResearchLLM Verifications Leaderboard

Data as of · 78 models, 80 questions · reviewed by Suhas Bhairav · methodology

Best LLM for…

The cheapest LLMs that score 75% or more

Answer: GPT-6.1 Sol by OpenAI: $0.0029 per question at a 76.2% score.

Reviewed by Suhas Bhairav · Data as of

Ranking

Models with a verification score of at least 75%, ordered by billed cost per question (reasoning tokens included). Top 15 of 17.

The cheapest LLMs that score 75% or more
#ModelCost / questionCost / correct answerScoreSpeed (tok/s)
1GPT-6.1 SolOpenAI$0.0029$0.003776.2%41
2Gemini 3.5 Flash LiteGoogle$0.0034$0.004575.0%409
3Gemini 3.8 FlashGoogle$0.0037$0.004582.5%142
4Muse Spark OpenMeta$0.0045$0.005877.5%220
5GPT-6.1 Terra ProOpenAI$0.005$0.006576.2%—
6Claude Sonnet 5.5Anthropic$0.0054$0.006978.8%126
7Gemini 4 ArgonGoogle$0.0062$0.007483.8%69
8GPT-5.6 TerraOpenAI$0.0063$0.008475.0%101
9GPT-5.3-CodexOpenAI$0.01$0.01382.5%95
10Kimi K3.1 ProMoonshotAI$0.01$0.01477.5%38
11Qwen3.8 Max (0902)Qwen$0.011$0.01575.0%46
12GPT-6 AstraOpenAI$0.013$0.01776.2%49
13Claude Opus 5.5Anthropic$0.015$0.01785.0%129
14Claude Fable 5.1Anthropic$0.016$0.0281.2%70
15Gemini 3.1 Pro PreviewGoogle$0.017$0.02277.5%124

How to read this

Scores come from 80 multiple-choice questions from SuperGPQA and MMLU-Pro. With that many questions, a score has a margin of error of several points, so models a few points apart are effectively tied on accuracy; the comparison pages run a significance test for any pair. To see the top two side by side, compare GPT-6.1 Sol and Gemini 3.5 Flash Lite.

Speed, latency and cost are measured through OpenRouter and depend on which host serves the request; your numbers on a direct API may differ. The full procedure is on the methodology page, and the data behind this ranking can be downloaded.

Other questions

Quick answers

What is the cheapest AI model that still scores 75% or more?
GPT-6.1 Sol by OpenAI: $0.0029 per question at a 76.2% score. Rule: Models with a verification score of at least 75%, ordered by billed cost per question (reasoning tokens included).
What is the runner-up?
Gemini 3.5 Flash Lite by Google: $0.0034 per question at a 75.0% score.
Which of these costs least per correct answer?
GPT-6.1 Sol: $0.0037 per correct answer (cost per question divided by its score).