1. Breaking down + understanding a long video I uploaded the entire NBA dunk contest from last night and asked which dunk had the highest score. Gemini 1.5 was incredibly able to find the specific perfect 50 dunk and details from just its long context video understanding!

Gemini 1.5 Pro's 1,000,000+ token context length is incredible. I got early access and spent my Saturday night running tests. Here are 6 impressive capabilities I found:

First, what is Gemini 1.5 Pro? Gemini 1.5 Pro is Google DeepMind's newest update to the free version of its LLM. It most notably has a 1M+ token context length, meaning it can understand much larger amounts of data compared to other LLMs/chatbots.

Gemini 1.5 Pro was overshadowed by OpenAI's Sora but is probably the biggest boost in LLM capabilities this year (so far). It's not available to the public yet, but Google DeepMind was kind enough to grant me early access. Let's get into the tests below.

2. Understanding and comparing full movie transcripts from Interstellar and Ad Astra. Gemini 1.5 was able to understand, compare, and contrast entire transcripts from both movies to help me decide which movie I should watch.

3. Translating language into a language spoken by less than 2000 people. Gemini 1.5 was able to translate English to Saterlandic (spoken by less than 2000 people) following a full linguistic manual at inference time. Absolutely incredible.

4. Watching, understanding, and distinguishing if an OpenAI Sora video is AI or not. Gemini 1.5 highlighted the famous Sora cat video and highlighted key factors of why it could be AI-generated. I was suprised by how in-depth it went on the response.

5. Finding, understanding, and explaining a small figure in a long paper. Gemini 1.5 was able to extract "table 8" from the Gemini 1.5 Pro paper by DeepMind and explain what the table meant. Again, note the details here.

6. Understanding the entire Interstellar movie transcript, and highlighting key moments. Gemini 1.5 was able to tell me the 3 most inspiring quotes from the transcript of Interstellar.

That's it for the initial testing. I'll be breaking down more Gemini tests, insights, and real-life implementations in my newsletter this week. Subscribe, and you'll never miss a thing in the world of AI ever again.

One last thing. I spent my entire Saturday night tinkering on Gemini 1.5 so I could give you this early content. If you found it valuable, spare me a follow @rowancheung for more deep AI insights.

Answering community prompts today with Gemini 1.5 Pro. Comment on the Tweet linked here and I'll try to get your prompt in:

I also saw it analyzing a 400 pages text book which is quite impressive

The ability to comprehend and comment on an hour of video within a few seconds of upload feels like another intellectual step forward for AI/Compute and another thing it can do much more capably than humans. The jump for language comprehension to video comprehension came a lot faster than I would have thought.