In 30 seconds
Matthew Berman calls GPT-6 Astra the best model he has used and points to three uses: generating a game, researching with browser control and writing. His short demonstrations and personal impressions are not a comparative benchmark.
- 1Make a gameBerman shows a playable Fall Guys-style game and says a single prompt can produce one.
- 2Research in a browserThe video shows browser control being used to research and compare rare Pokémon cards.
- 3Try its writingBerman says its writing has less of an AI feel than that of other models, although some remains.
Takeaways
- Berman shows a Fall Guys-style game that he says was created with a single prompt.
- He demonstrates browser research comparing rare Pokémon cards.
- He considers its writing less recognizably AI-generated than other models, while acknowledging some remains.
Read the edited text by chapter
A strong first impression
Berman describes GPT-6 Astra as the best model he has used. He suggests three things to try.
GPT-6 Astra is the best model I have ever used. Here are three things I recommend trying with it: making games, controlling a browser to do research, and writing.
Game creation
He shows a Fall Guys-style game and says it was made with a single prompt. This is a brief demonstration rather than a systematic test.
First, create a game. I show a playable Fall Guys-style game and say that a single prompt can produce something fun. This short example is a demonstration, not evidence that every one-prompt game will work equally well.
Browser control
The model is shown researching rare Pokémon cards. Berman suggests using browser control for practical work.
Next, use browser control to do practical work. In the example, the model quickly researches and compares rare Pokémon cards. The clip shows a research task, but does not establish the completeness or accuracy of its findings.
Writing quality
Berman offers a favorable but subjective comparison with other models. He acknowledges that the writing can still sound AI-generated.
Finally, try writing with it. In my experience, it is better than other models and has less of that recognizable AI feel, though some remains. That is my impression rather than a controlled comparison; try it and decide what you think.
More from this channel
34:21EN
Matthew Berman· Oct 27, 2026Matthew 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.
0:36ENMatthew Berman· Oct 25, 2026Kimi 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.
25:50ENMatthew Berman· Oct 24, 2026Kimi 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.
32:20ENMatthew Berman· Oct 23, 2026Who 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.
1:03:37ENMatthew Berman· Oct 23, 2026Matthew Berman’s Team Tests Kimi K3.1 on Animation, Games, and ChartsMatthew Berman, Alex, Brian · Forward FutureA live team AMA turns into an informal test of Kimi K3.1, from personal-agent workflows to motion graphics, a browser game, and a benchmark chart.
0:53ENMatthew Berman· Oct 22, 2026Sam Altman on AI and the Discovery of New KnowledgeSam Altman · OpenAIAltman expects progress in mathematical discovery to extend to biology, materials and energy.