Matthew Berman Tests a Gemini 4 Argon Fall Guys-Style Game

A short playthrough shows a colorful obstacle course after a few rounds of iteration, with Berman assessing its gameplay and collisions.

MBMatthew Berman登壇者 · Forward Future
チャンネル
Matthew Berman
公開日
2026年10月17日
動画
0:42
読了時間
1 分
スライド
3
原文
英語
YouTube で開く
スクロールすると動画は右下に小さくなって再生を続けます。字幕は閲覧言語に合わせて切り替わります。
閲覧言語2 言語で読めます
この記事はまだ日本語に翻訳されていないため、英語の原文を表示しています。
要約

30 秒でわかる

Matthew Berman plays a Fall Guys-style game that the video attributes to Gemini 4 Argon. He says he iterated a couple of times, praises the level and character collisions, and finds the course unexpectedly difficult before reaching the next stage and apparently finishing first. The clip does not show the prompts, code, full development process or a controlled comparison with other models.

  1. 1A familiar game as a testBerman tries a Fall Guys-style clone, which the video title attributes to Gemini 4 Argon. His view that such games represent the limit for frontier models is an informal judgment.
  2. 2A few iterationsHe says he revised the game only a couple of times and is pleased with its level and collisions.
  3. 3A difficult courseAs obstacles bounce his character around, he calls this the hardest version he has created so far.
  4. 4Reaching the next stageHe reaches another stage and says it looks like he finishes first.

ポイント

  1. Berman says he iterated on the game a couple of times before the playthrough.
  2. The demonstration shows jumping, obstacles and collisions among bean-shaped characters.
  3. Berman finds the level fun and harder than earlier versions he had made.
  4. The short gameplay clip does not establish the full development effort or model performance across games.
全文

チャプターごとに整えた全文を読む

文字起こしのみ
01

Playing the generated obstacle course

Berman introduces the Fall Guys-style game and says he iterated on it a couple of times. He likes its look, jumping and character collisions.

Players begin the obstacle course
スライド 1 / 3Players begin the obstacle courseBean-shaped characters start a colorful course with moving obstacles.

I wanted to try a Fall Guys-style clone because, in my experience, that kind of game seems close to the limit of what frontier models can create. The video identifies this version with Google’s Gemini 4 Argon. This is a brief example rather than a controlled comparison or proof of a general limit.

Let’s play. The level looks good, and I think I iterated on it only a couple of times. The character jumps, the little bean-shaped characters collide, and the levels are actually fun. The clip does not show the prompts, code or all of the work behind the game.

I am clearly not very good at it, and the course turns out to be harder than I thought. As I go up the slide, obstacles bounce me around; this is actually the hardest version I have created so far.

02

A difficult route to the finish

Obstacles make the level harder than he expected. He reaches another stage and thinks he finishes first.

Navigating the obstacles
スライド 2 / 3Navigating the obstaclesThe player encounters the bouncing obstacles that Berman says make the level challenging.
Another stage of the course
スライド 3 / 3Another stage of the courseThe playthrough moves on to rotating platforms on the way toward the finish.

I make it to the next stage and try to reach the finish line. It looks as though I got first place. The playthrough shows a working moment in this particular build, not whether every feature or game mode works reliably.

同じチャンネルの記事

スライド 12 枚34:21EN
Matthew Berman· 2026年10月27日Matthew Berman Examines Anthropic’s IPO Filing, AI Growth and GovernanceMatthew Berman, Sam Altman · Forward FutureThe presenter reviews the S-1’s financial metrics, compute strategy and voting structure while questioning the tension between rapid AI development and safety.AnthropicAI Safetyエンタープライズ AI+2
0:26EN
Matthew Berman· 2026年10月25日Sam Altman on Human Values and Problem-Solving in an AI WorldSam Altman · OpenAIIn a short interview excerpt, Altman considers the enjoyment of solving hard problems and the human role in interpreting results.Human-AI CollaborationFuture of WorkAI Strategy
スライド 2 枚0:36EN
Matthew Berman· 2026年10月25日Kimi K3.1 and a Rubik’s Cube Solver DemonstrationMatthew Berman · Forward FutureMatthew Berman shows a scrambled virtual cube being solved and gives the Kimi K3.1 test a favorable assessment.評価(Evals)コーディングエージェントOpen-weight models
スライド 2 枚0:21EN
Matthew Berman· 2026年10月24日An AI-Generated Kimi Benchmark Chart Used Invented ScoresMatthew Berman · Forward FutureMatthew Berman asks for a graph of Kimi K3.1 Pro benchmarks, then discovers that the displayed figures are illustrative placeholders.Benchmarks評価(Evals)Generative media
スライド 6 枚25:50EN
Matthew Berman· 2026年10月24日Kimi K3.1 Pro Tested: Strong 3D Demos, Uneven Results and Competitive PricingMatthew Berman · Forward FutureMatthew Berman tests the open-weight model on coding, 3D rendering and visual explanations, then compares benchmark results and prices.Open-weight modelsBenchmarksAI cost & performance+1
スライド 2 枚32:20EN
Matthew Berman· 2026年10月23日Who Is Local AI For? The Costs, Trade-offs and a Hybrid FutureMatthew Berman · Forward FutureMatthew Berman weighs data control and independence against hardware cost, setup complexity and inference speed.Local inferenceOpen-weight modelsAI cost & performance+1