Mysterious AI from Food Delivery App
40sThe anonymous origin story from a non-tech giant creates curiosity and shares well as a surprising underdog narrative.
▶ Play ClipAI Summary
Long Cap 2.0 is a new free and open-source Chinese AI model developed by a food delivery company, not a tech giant. It features 1.6 trillion parameters with Mixture of Experts, a 1 million token context window, and outperforms GPT-5.5 on the SWE-bench pro benchmark.
Chapters
[00:00]
Origin
Developed by a food delivery app, not a major tech company.
[00:00]
Anonymous Testing
Ran anonymously on open router for 2 months under the codename Alpha.
[00:00]
Model Architecture
1.6 trillion parameters with Mixture of Experts, activating only needed parameters per task.
[00:00]
Context Window
1 million tokens of context, allowing entire knowledge bases to be processed at once.
[00:00]
Performance
Scored 59.5 on SWE-bench pro, beating GPT-5.5 which scored 58.6.
[00:00]
License
Fully open-source under the MIT license, free for anyone to use and build upon.
Long Cap 2.0 demonstrates that impressive AI advancements can come from unexpected sources, offering competitive performance and full open-source accessibility.
Study Flashcards (6)
What is the context length of Long Cap 2.0?
easy
Click to reveal answer
What is the context length of Long Cap 2.0?
easy
Click to reveal answer
1 million tokens.
How many parameters does Long Cap 2.0 have?
easy
Click to reveal answer
How many parameters does Long Cap 2.0 have?
easy
Click to reveal answer
1.6 trillion parameters.
What technique does Long Cap 2.0 use to save power?
medium
Click to reveal answer
What technique does Long Cap 2.0 use to save power?
medium
Click to reveal answer
Mixture of Experts, activating only needed parameters for each task.
What was the codename of Long Cap 2.0 during anonymous testing?
easy
Click to reveal answer
What was the codename of Long Cap 2.0 during anonymous testing?
easy
Click to reveal answer
Alpha.
How did Long Cap 2.0 perform on SWE-bench pro compared to GPT-5.5?
medium
Click to reveal answer
How did Long Cap 2.0 perform on SWE-bench pro compared to GPT-5.5?
medium
Click to reveal answer
It scored 59.5, beating GPT-5.5's 58.6.
Under what license is Long Cap 2.0 released?
easy
Click to reveal answer
Under what license is Long Cap 2.0 released?
easy
Click to reveal answer
MIT license.
💡 Key Takeaways
📊
Unexpected Origin
The model was built by a food delivery app, not a tech giant, challenging assumptions about AI development.
💡
Anonymous Success
It climbed leaderboards under a codename without revealing its creators.
🔧
Massive Context Window
1 million token context is a significant leap, enabling processing of entire knowledge bases.
📊
Outperforming GPT
Beating GPT-5.5 on SWE-bench pro shows competitive performance from a smaller team.
⚖️
Full Open Source
MIT license allows unrestricted use and modification, fostering community development.
Full Transcript
New free, plus open-source Chinese AI is insane. This new Chinese AI has 1 million tokens of memory for free. This new Chinese AI is called Long Cap 2.0, and it just came from a food delivery app, not a tech giant. It ran anonymously on open router for 2 months under the code name Alpha, climbing the charts with nobody knowing who built it. It has 1.6 trillion parameters, but only activates the ones it needs for each task,
saving power and running faster. The real game-changer is 1 million tokens of context. Feed it an entire knowledge base, and it holds all of it in memory at once. On SWE bench pro, it scored 59.5, beating GPT 5.5 at 58.6, and it's fully open-source under the MIT license, free for anyone to build on.
