AI trivia questions with answers
49 AI trivia questions for quiz nights, classrooms or just testing yourself. Tap a question to see the answer and its source. The second half follows today’s models, so it changes as new ones come out.
AI history and basics
2.Which 2017 research paper introduced the transformer, the design behind today’s chat models?Show answer
“Attention Is All You Need”, by a team at Google.
Source: arXiv
4.Who proposed the “imitation game”, now called the Turing test, and in what year?Show answer
Alan Turing, in his 1950 paper “Computing Machinery and Intelligence”.
Source: Wikipedia
5.Where was the term “artificial intelligence” first used?Show answer
In the 1955 proposal for the 1956 Dartmouth summer workshop, written by John McCarthy, Marvin Minsky, Nathaniel Rochester and Claude Shannon.
Source: Wikipedia
6.What was the name of the 1960s MIT chatbot that played a psychotherapist?Show answer
ELIZA, written by Joseph Weizenbaum.
Source: Wikipedia
7.Which IBM computer beat world chess champion Garry Kasparov in a match in 1997?Show answer
Deep Blue.
Source: Wikipedia
8.Which program beat Go champion Lee Sedol four games to one in 2016?Show answer
AlphaGo, from DeepMind.
Source: Wikipedia
9.Which IBM system beat Ken Jennings and Brad Rutter on Jeopardy! in 2011?Show answer
Watson.
Source: Wikipedia
11.Who shared the 2024 Nobel Prize in Physics for foundational work on neural networks?Show answer
John Hopfield and Geoffrey Hinton.
Source: Nobel Prize
12.Which AI system earned Demis Hassabis and John Jumper half of the 2024 Nobel Prize in Chemistry?Show answer
AlphaFold, which predicts the 3D shape of proteins. David Baker won the other half for protein design.
Source: Nobel Prize
14.Roughly how much English text is one token?Show answer
About three quarters of a word: 100 tokens is around 75 words.
Source: OpenAI Help Center
15.What was the LMArena leaderboard called before 2025?Show answer
Chatbot Arena. How it ranks models.
Source: LMArena
16.What does the “70B” in a model name like Llama 3.3 70B mean?Show answer
70 billion parameters. What parameters are.
Source: Wikipedia
17.What is a model’s context window?Show answer
How much text it can read at once, counted in tokens. More on context windows.
Source: Wikipedia
18.What does it mean when a model has “open weights”?Show answer
You can download the trained model and run it yourself. That is not always open source: the training code and data often stay private. Open weights vs closed models.
Source: Wikipedia
19.Who founded Anthropic, the company behind Claude?Show answer
Former OpenAI staff, including the siblings Dario and Daniela Amodei, in 2021.
Source: Wikipedia
Today’s AI models
25.Which company makes DeepSeek V4.1 Flash?Show answer
DeepSeek.
Source: DeepSeek V4.1 Flash on Modeldle
29.Which can read more text at once: GPT-6.1 Sol or Claude Opus 5.5?Show answer
GPT-6.1 Sol: 1.05M tokens, against 1M tokens for Claude Opus 5.5.
Source: OpenAI docs
30.Which can read more text at once: GPT-6.1 Sol or Grok 4.7?Show answer
GPT-6.1 Sol: 1.05M tokens, against 500K tokens for Grok 4.7.
Source: OpenAI docs
31.Which can read more text at once: Claude Opus 5.5 or Grok 4.7?Show answer
Claude Opus 5.5: 1M tokens, against 500K tokens for Grok 4.7.
Source: Anthropic docs
32.Which can read more text at once: GPT-6.1 Sol or DeepSeek V4.1 Flash?Show answer
GPT-6.1 Sol: 1.05M tokens, against 1M tokens for DeepSeek V4.1 Flash.
Source: OpenAI docs
33.Which costs less to use through its API: Claude Opus 5.5 or DeepSeek V4.1 Flash?Show answer
DeepSeek V4.1 Flash: $0.15 in and $0.60 out per million tokens, against $4.00 and $20.00 for Claude Opus 5.5.
Source: DeepSeek docs
34.Which costs less to use through its API: DeepSeek V4.1 Flash or Qwen3.8 Max?Show answer
DeepSeek V4.1 Flash: $0.15 in and $0.60 out per million tokens, against $2.00 and $6.00 for Qwen3.8 Max.
Source: DeepSeek docs
35.Which costs less to use through its API: DeepSeek V4.1 Flash or Kimi K3.1 Pro?Show answer
DeepSeek V4.1 Flash: $0.15 in and $0.60 out per million tokens, against $2.00 and $10.00 for Kimi K3.1 Pro.
Source: DeepSeek docs
36.Which costs less to use through its API: Qwen3.8 Max or Kimi K3.1 Pro?Show answer
Qwen3.8 Max: $2.00 in and $6.00 out per million tokens, against $2.00 and $10.00 for Kimi K3.1 Pro.
Source: Alibaba docs
37.Which costs less to use through its API: Muse Wonder 1.0 or Qwen3.8 Max?Show answer
Qwen3.8 Max: $2.00 in and $6.00 out per million tokens, against $4.00 and $20.00 for Muse Wonder 1.0.
Source: Alibaba docs
38.Which ranks higher on the LMArena text leaderboard: GPT-6.1 Sol or Claude Opus 5.5?Show answer
Claude Opus 5.5, 1514 points to 1445. People prefer its answers in about 60% of blind votes.
Source: LMArena text leaderboard
39.Which ranks higher on the LMArena text leaderboard: GPT-6.1 Sol or Gemini 4 Carbon?Show answer
Gemini 4 Carbon, 1538 points to 1445. People prefer its answers in about 63% of blind votes.
Source: LMArena text leaderboard
40.Which ranks higher on the LMArena text leaderboard: Claude Opus 5.5 or Gemini 4 Carbon?Show answer
Gemini 4 Carbon, 1538 points to 1514. People prefer its answers in about 53% of blind votes.
Source: LMArena text leaderboard
41.Which ranks higher on the LMArena text leaderboard: GPT-6.1 Sol or Grok 4.7?Show answer
GPT-6.1 Sol, 1445 points to 1400. People prefer its answers in about 56% of blind votes.
Source: LMArena text leaderboard
42.Which ranks higher on the LMArena text leaderboard: Gemini 4 Carbon or Grok 4.7?Show answer
Gemini 4 Carbon, 1538 points to 1400. People prefer its answers in about 69% of blind votes.
Source: LMArena text leaderboard
43.Can you download DeepSeek V4 Pro and run it on your own hardware?Show answer
Yes: DeepSeek publishes its weights.
Source: DeepSeek docs
44.Can you download GPT-6.1 Sol and run it on your own hardware?Show answer
No. OpenAI offers it only through its app and API.
Source: OpenAI docs
45.Can you download DeepSeek V4 Flash and run it on your own hardware?Show answer
Yes: DeepSeek publishes its weights.
Source: DeepSeek docs
46.Can you download Claude Opus 5.5 and run it on your own hardware?Show answer
No. Anthropic offers it only through its app and API.
Source: Anthropic docs
47.Which image model tops the LMArena text-to-image leaderboard?Show answer
GPT Image 2.5, with 1429 points. See the full ranking.
48.Which video model tops the LMArena text-to-video leaderboard?Show answer
Gemini Omni 1.2 Flash, with 1540 points. See the full ranking.
49.Which coding agent has the best Terminal-Bench result?Show answer
Claude Code, with 64.9% of tasks. See the full ranking.
Source: Terminal-Bench 4.0 leaderboard
Want more? Today’s Modeldle quiz asks eight new questions about real AI models every day, and the comparison tool puts any models side by side.