正在加载视频...

视频加载失败

加载此视频时出现问题。这可能是由于临时网络问题，或视频可能不可用。

OpenAI released o3 and o4-mini! I've tested both against the top reasoning models using the exact same prompt

Flavio Adamo

9,582 subscribers

262,286 次观看 • 1 年前 •via X (Twitter)

艺术游戏科学技术

Anya Rossi• Live Now

Private livecam show

11 条评论

Rob Haisfield 的头像

Rob Haisfield1 年前

Seems like the benchmark got saturated!

Flavio Adamo 的头像

Flavio Adamo1 年前

I’m working on something 👀

Ithy 的头像

Ithy1 年前

What happens when you combine every AI? It's time for something better than ChatGPT...

Dan 的头像

Dan1 年前

we need a more complex prompt. have it make a rube goldberg machine

Flavio Adamo 的头像

Flavio Adamo1 年前

Working on it!

Mason Flicek 的头像

Mason Flicek1 年前

o3 might be AGI

Anton P. 🇮🇹 的头像

Anton P. 🇮🇹1 年前

Love it, Flavio! Keep hexagonning :)

Harmony Hacker 的头像

Harmony Hacker1 年前

DeepSeek R2 will need to put on a show. Anything that beats 2.5, o3, and o4 going forward should be classified as AGI.

Robert 的头像

Robert1 年前

Sorry, how do I know what's good from looking at these things? They all seem to work. Doesn't seem like a very decisive test

NewAIWorld 的头像

NewAIWorld1 年前

o3 is really something else 🥰

Eis Maus 的头像

Eis Maus1 年前

Interesting to note that Gemini nails the ball interactions, and that none of the other models do. Share the prompt you used?

相关视频

OpenAI just dropped 3 new models! I've tested them all against 4.5 with the exact same prompt

OpenAI just dropped 3 new models! I've tested them all against 4.5 with the exact same prompt

Flavio Adamo

303,424 次观看 • 1 年前

Introducing OpenAI o3 and o4-mini—our smartest and most capable models to date. For the first time, our reasoning models can agentically use and combine every tool within ChatGPT, including web search, Python, image analysis, file interpretation, and image generation.

Introducing OpenAI o3 and o4-mini—our smartest and most capable models to date. For the first time, our reasoning models can agentically use and combine every tool within ChatGPT, including web search, Python, image analysis, file interpretation, and image generation.

OpenAI

3,720,579 次观看 • 1 年前

You all saw that OpenAI Super Bowl ad, right? Well, I recreated the exact same effect using ONLY OpenAI tools Code - o3-mini Video - Sora Try it here

You all saw that OpenAI Super Bowl ad, right? Well, I recreated the exact same effect using ONLY OpenAI tools Code - o3-mini Video - Sora Try it here

Mike Bespalov

412,129 次观看 • 1 年前

Web search is now available with OpenAI o3, o3-pro, and o4-mini. The model can search the web within its chain-of-thought! 🧠🌐 $10 / 1K tool calls

Web search is now available with OpenAI o3, o3-pro, and o4-mini. The model can search the web within its chain-of-thought! 🧠🌐 $10 / 1K tool calls

OpenAI Developers

190,186 次观看 • 1 年前

This week, we tested 3 latest models in our Game Arena Benchmark: → O3 → O4-mini → Gemini 2.5 Flash Across 4 games—Phoenix Wright, Sokoban, Candy Crush, and 2048—O3 dominated the zero-shot leaderboard, ranking #1 or #2 in nearly every task and outperforming previous SOTA models like O3-mini and Gemini 2.5 Flash. 🔥 Beyond our customized tests, we took on a real challenge: Sokoban (1989)—the classic, unforgiving original. 🗣️ Many say O3 shows strong image reasoning, but does it really? Let’s see what our Game Arena Benchmark reveals. 👇

This week, we tested 3 latest models in our Game Arena Benchmark: → O3 → O4-mini → Gemini 2.5 Flash Across 4 games—Phoenix Wright, Sokoban, Candy Crush, and 2048—O3 dominated the zero-shot leaderboard, ranking #1 or #2 in nearly every task and outperforming previous SOTA models like O3-mini and Gemini 2.5 Flash. 🔥 Beyond our customized tests, we took on a real challenge: Sokoban (1989)—the classic, unforgiving original. 🗣️ Many say O3 shows strong image reasoning, but does it really? Let’s see what our Game Arena Benchmark reveals. 👇

Hao AI Lab

14,636 次观看 • 1 年前

It's finally here. OpenAI just dropped o3 and o4-mini. People can’t believe how "intelligent" it is. 10 wild examples: 👇

It's finally here. OpenAI just dropped o3 and o4-mini. People can’t believe how "intelligent" it is. 10 wild examples: 👇

Min Choi

231,597 次观看 • 1 年前

It appears that the exact same prompt was used to generate both the anti-Ukraine and anti-Israel AI videos, both created using Veo. Both videos share the exact same length (8 seconds) and nearly identical framing. See here:

It appears that the exact same prompt was used to generate both the anti-Ukraine and anti-Israel AI videos, both created using Veo. Both videos share the exact same length (8 seconds) and nearly identical framing. See here:

Tal Hagin

17,378 次观看 • 1 年前

🔴 ¡OPENAI ANUNCIA o3 y o4-mini! 🔴 La saga de modelos razonadores se actualiza con dos nuevos modelos: o3 y o4-mini. Estos modelos pasarán a sustituir a los modelos o1 al considerarse superiores en todas sus capacidades. Os cuento 👇

🔴 ¡OPENAI ANUNCIA o3 y o4-mini! 🔴 La saga de modelos razonadores se actualiza con dos nuevos modelos: o3 y o4-mini. Estos modelos pasarán a sustituir a los modelos o1 al considerarse superiores en todas sus capacidades. Os cuento 👇

Carlos Santana

53,112 次观看 • 1 年前

Update on the new reasoning popover in ChatGPT web app prompt composer - there's now even a keyboard shortcut to cycle through reasoning levels, and it looks like these levels correspond to "Quick" (low) = GPT-4o, "Think a little" (medium) = o3-mini, and "Think harder" (high) = o3-mini-high

Update on the new reasoning popover in ChatGPT web app prompt composer - there's now even a keyboard shortcut to cycle through reasoning levels, and it looks like these levels correspond to "Quick" (low) = GPT-4o, "Think a little" (medium) = o3-mini, and "Think harder" (high) = o3-mini-high

Tibor Blaho

63,219 次观看 • 1 年前

SEEDANCE 2 vs FAST vs MINI Seedance 2.0 mini just went live on CapCut and I tested the exact same storyboard for reference and prompt to see which one performs best. And the difference is actually crazy 😳 Which one do you think came out on top? 👀 #CapcutSeedance2

SEEDANCE 2 vs FAST vs MINI Seedance 2.0 mini just went live on CapCut and I tested the exact same storyboard for reference and prompt to see which one performs best. And the difference is actually crazy 😳 Which one do you think came out on top? 👀 #CapcutSeedance2

MrDejie

133,840 次观看 • 25 天前

DeepSeek R1 671B just broke speed records at 198 t/s it is now the fastest reasoning model available you can try it in coding mode on anychat soon prompt: make two fancy D3.js animated looping speedometers, first one is title deepseek R1 671B and second one is OpenAI o3-mini, deep seek should go up to 198 tokens/second and openai o3-mini should go up to 174 tokens/second

DeepSeek R1 671B just broke speed records at 198 t/s it is now the fastest reasoning model available you can try it in coding mode on anychat soon prompt: make two fancy D3.js animated looping speedometers, first one is title deepseek R1 671B and second one is OpenAI o3-mini, deep seek should go up to 198 tokens/second and openai o3-mini should go up to 174 tokens/second

AK

135,360 次观看 • 1 年前

Seedance 2.0 vs. Kling 3.0: Who delivers the most realistic emotional expressions? I put both models through the same test using one character, one reference image, and the exact same prompt. Each generated four emotions: anger, sadness, happiness, and disgust. Which AI video model handled the emotions better ?

Seedance 2.0 vs. Kling 3.0: Who delivers the most realistic emotional expressions? I put both models through the same test using one character, one reference image, and the exact same prompt. Each generated four emotions: anger, sadness, happiness, and disgust. Which AI video model handled the emotions better ?

FATHELA ESQ

38,601 次观看 • 12 天前

Today, we shared evals for an early version of the next model in our o-model reasoning series: OpenAI o3

Today, we shared evals for an early version of the next model in our o-model reasoning series: OpenAI o3

OpenAI

1,668,850 次观看 • 1 年前

Greg Brockman says o3 and o4-mini mark "a qualitative step into the future." These models don’t just reason -- they act. One used 600 tool calls in a row to solve a complex task. "Top scientists tell us they produce legitimately good and useful novel ideas."

Greg Brockman says o3 and o4-mini mark "a qualitative step into the future." These models don’t just reason -- they act. One used 600 tool calls in a row to solve a complex task. "Top scientists tell us they produce legitimately good and useful novel ideas."

vitrupo

96,514 次观看 • 1 年前

I've long said that o3 is the best coding model - but if you're using an agent harness - Claude is just better at navigating your codebase. Enter the Repo Prompt pair programmer mode - it's the best of both words, as Claude coordinates with o3 to plan and apply edits for you!

I've long said that o3 is the best coding model - but if you're using an agent harness - Claude is just better at navigating your codebase. Enter the Repo Prompt pair programmer mode - it's the best of both words, as Claude coordinates with o3 to plan and apply edits for you!

eric provencher

93,571 次观看 • 11 个月前

holy shit Kimi K2.6 literally cooked claude design🤯 and it's 7x cheaper try the same prompt on both Kimi and Claude design, and there are the outputs see the exact same prompt 👇

holy shit Kimi K2.6 literally cooked claude design🤯 and it's 7x cheaper try the same prompt on both Kimi and Claude design, and there are the outputs see the exact same prompt 👇

Guri Singh

303,140 次观看 • 2 个月前