I Tried LongCat 2.0 and It Is Incredible (Open Source)
A hands-on gauntlet of LongCat 2.0 — the 1.6-trillion-parameter MIT-licensed MoE model from a Chinese food-delivery company that ran anonymously on OpenRouter for two months — covering three coding and data demos run in a sandbox, its painful China-only signup and Alipay billing, promo pricing of $0.30/$1.20 per million tokens, and an architect-executor workflow pairing Fable 5 planning with LongCat execution.
Kacper Rutkiewicz | AI Made Simple20 minTranscript found
Quick learning frame
Read this before watching.
Creative automation uses agents to accelerate production while keeping human taste in story, pacing, selection, and critique.
New playlist item from Kacper Rutkiewicz | AI Made Simple; queued for transcript-backed review, topic mapping, and a practical learning artifact.
Skill you build: The ability to evaluate a cheap open-source model on real agentic tasks and deploy it as the low-cost executor in an architect-executor orchestration where a frontier model does the planning and review.
Watch for the shift from claim to mechanism. The learning value is the point where the transcript reveals a repeatable action, tool boundary, context move, review habit, or artifact.
Concept diagram
Where this video fits.
01Brief
02Source
03Generation
04Selection
05Edit
06Taste Review
Deep lesson
Turn this video into working knowledge.
5,443 cleaned transcript words reviewed across 1,466 timed caption segments.
Thesis
I Tried LongCat 2.0 and It Is Incredible (Open Source) teaches a practical creative automation move: A hands-on gauntlet of LongCat 2.0 — the 1.6-trillion-parameter MIT-licensed MoE model from a Chinese food-delivery company that ran anonymously on OpenRouter for two months — covering three coding and data demos run in a sandbox, its painful China-only signup and Alipay billing, promo pricing of $0.30/$1.20 per million tokens, and an architect-executor workflow pairing Fable 5 planning with LongCat execution.
The goal is not to remember the video. The goal is to extract the operating principle, tie it to timestamped evidence, test how far the claim transfers, and make something reusable.
0:38
The anonymous OpenRouter hit
“we're going to be running. We're going to build an app, which is a full CLI tool from one description. This is going to show off the agentic coding of this model. Then, we're going to create a...”
A Chinese food-delivery company open-sourced the 1.6-trillion-parameter LongCat 2.0 after it sat on OpenRouter for two months under an anonymous name and climbed to global top-three daily call volume purely on quality; because access is API-only through a Chinese platform, the review runs every demo inside a sandbox so no personal or client data travels to China. Before testing any API-only model, write down exactly what data would transit the provider and set up a sandbox or throwaway project so nothing sensitive leaves your machine.
6:20
Fast, cheap, loop-hungry
“if you prompt the model properly and give it kind of verification points and specific tasks that it needs to complete, it will loop until it gets that done, and then the results actually become very solid. This...”
The demos — a CLI expense tracker planned and tested in under two minutes for under a cent, a 30-second 3D solar system, and a plain-English SQLite revenue query — show LongCat is extremely fast and dirt cheap (about 100K tokens and under five cents for everything), but it is only okay at one-shot prompting: it is built to loop as an agent against verification points, a worker rather than a search box. Rerun one small build task twice — once as a single one-shot prompt, once with explicit verification checkpoints for the model to loop against — and compare the output quality.
14:35
Architect-executor orchestration
“want to do the best job reviewing them that I can, but I genuinely like this model. Okay, so there we go. Our Pi agent finished up. All the fixes were applied cleanly. You guys can see the...”
Pairing Fable 5 on high effort as the architect (writing the PRD, then review fixes) with LongCat 2.0 in Pi agent as the executor produced an eight-section landing page in about 90 seconds for under two cents, with a second loop fixing the broken anchor navigation and waitlist form Fable flagged — let the expensive model think and the cheap model do the work at a 20–40x lower execution cost. Have a frontier model write a tight PRD with success criteria for a one-page site, hand it to a cheap model to build, then feed the frontier model's review notes back as a fixes file for another execution loop.
01
Brief
Start with this video's job: A hands-on gauntlet of LongCat 2.0 — the 1.6-trillion-parameter MIT-licensed MoE model from a Chinese food-delivery company that ran anonymously on OpenRouter for two months — covering three coding and data demos run in a sandbox, its painful China-only signup and Alipay billing, promo pricing of $0.30/$1.20 per million tokens, and an architect-executor workflow pairing Fable 5 planning with LongCat execution. Treat "Brief" as the outcome you are trying to make visible, not a topic label. Anchor it to 0:38, where the video says: “we're going to be running. We're going to build an app, which is a full CLI tool from one description. This is going to show off the agentic coding of this model. Then, we're going to create a...”
02
Source
Use "Source" to locate the part of the creative automation workflow the video is demonstrating. Ask what changes in your real setup if this claim is true. Anchor it to 6:20, where the video says: “if you prompt the model properly and give it kind of verification points and specific tasks that it needs to complete, it will loop until it gets that done, and then the results actually become very solid. This...”
03
Generation
Turn "Generation" into the reusable artifact for this lesson: A creative workflow board with critique criteria and review checkpoints. This is where watching becomes something you can inspect and reuse.
04
Selection
Use "Selection" as the application surface. Decide whether the idea touches a browser flow, a local file, a model choice, a source document, a UI, or a review step.
05
Edit
Use "Edit" to prove the lesson. The evidence should connect back to the video title, transcript anchors, and a concrete output, not a generic best-practice claim.
06
Taste Review
Use "Taste Review" to carry the idea forward: save the prompt, checklist, diagram, or operating rule that would make the next agent run better.
Example
Source-backed work packet
Convert the video into a scoped task that includes the transcript claim, target workflow, acceptance criteria, and proof. The output should be a creative workflow board with critique criteria and review checkpoints..
Example
Claim vs. demo brief
Separate what the speaker claims, what the demo actually proves, and what still needs outside verification before you adopt the workflow.
Example
Teach-back module
Transform the lesson into a definition, a mechanism diagram, one misconception, one practice exercise, and a check-for-understanding question.
Do not learn it wrong
Treating the title as the lesson without checking what the transcript actually says.
Letting the prompt drift into generic advice that could apply to any video in the playlist.
Copying the tool setup without identifying the operating principle that transfers to your own stack.
Skipping the artifact, which means the learning never becomes operational or inspectable.
Do not count this as learned until these are true.
01
State the transcript-backed claim in your own words: A hands-on gauntlet of LongCat 2.0 — the 1.6-trillion-parameter MIT-licensed MoE model from a Chinese food-delivery company that ran anonymously on OpenRouter for two months — covering three coding and data demos run in a sandbox, its painful China-only signup and Alipay billing, promo pricing of $0.30/$1.20 per million tokens, and an architect-executor workflow pairing Fable 5 planning with LongCat execution.
02
Explain the practical stakes without hype: New playlist item from Kacper Rutkiewicz | AI Made Simple; queued for transcript-backed review, topic mapping, and a practical learning artifact.
03
Map the idea onto the Brief -> Source -> Generation -> Selection -> Edit -> Taste Review sequence and name the weakest link.
04
Produce the artifact and include the evidence that proves it: A creative workflow board with critique criteria and review checkpoints.
Put it into practice
Give this grounded prompt to Codex or Claude after watching.
You are helping me turn one specific YouTube video into real, durable learning.
Source video:
- Title: I Tried LongCat 2.0 and It Is Incredible (Open Source)
- URL: https://www.youtube.com/watch?v=P88eG0Bdduo
- Topic: Creative Automation
- My current learning frame: Run your own architect-executor loop: a frontier model writes a PRD with success criteria, LongCat 2.0 or another cheap model builds it, the frontier model reviews the output, the cheap model applies the fixes — then total the token bill for the entire cycle.
- Why this matters: New playlist item from Kacper Rutkiewicz | AI Made Simple; queued for transcript-backed review, topic mapping, and a practical learning artifact.
Transcript anchors from this exact video:
- 0:38 / Evidence 1: "we're going to be running. We're going to build an app, which is a full CLI tool from one description. This is going to show off the agentic coding of this model. Then, we're going to create a..."
- 2:59 / Evidence 2: "system. Super simple concept. I just want to show you guys the visual part of this model. So, let's go ahead and send this prompt off, and then we'll evaluate the results together. I'm not going to lie..."
- 6:20 / Evidence 3: "if you prompt the model properly and give it kind of verification points and specific tasks that it needs to complete, it will loop until it gets that done, and then the results actually become very solid. This..."
- 11:45 / Evidence 4: "to have Claude Fable 5 on the left-hand side here with high effort and then our Pi agent with medium effort on LongCat 2.0. Because Fable 5 is so good, all I'm going to do in natural language..."
- 14:35 / Evidence 5: "want to do the best job reviewing them that I can, but I genuinely like this model. Okay, so there we go. Our Pi agent finished up. All the fixes were applied cleanly. You guys can see the..."
- 16:10 / Evidence 6: "cents making this entire video. It's fully open-source. It's got an open MIT license. It's a real 1 million token context window. And it has some SMB uses beyond code. Some of the asterisks and weak points is..."
- 17:53 / Evidence 7: "free resource guide inside of my free school community AI Automation Nexus. All you need to do is click the link down below, enter in your email, join the community, you'll get access to all the free resource..."
Your task:
1. Use the transcript anchors above as the primary source packet. If you add outside context, label it clearly as outside context and keep it secondary.
2. Create a source-check table with columns: timestamp, claim, what the demo proves, confidence, and what still needs verification.
3. Extract the actual teachable claims from the video. Do not invent claims that are not supported by the title, lesson frame, or transcript anchors.
4. Build a reusable learning artifact: A creative workflow board with critique criteria and review checkpoints.
5. Include:
- a plain-English definition of the core idea
- a diagram or structured model using this sequence: Brief -> Source -> Generation -> Selection -> Edit -> Taste Review
- 3 concrete examples that apply the video idea to real agentic work
- 2 failure modes the video helps prevent
- a checklist I can use the next time I run Codex or Claude
- one practical exercise with a clear done signal
6. Add a "learning transfer" section: what changes in my workflow tomorrow if I actually learned this?
7. Add a "source check" section that cites which transcript anchor supports each major takeaway.
Quality bar:
- Make this specific to "I Tried LongCat 2.0 and It Is Incredible (Open Source)", not a generic Creative Automation essay.
- Prefer operational examples, failure modes, and reusable artifacts over broad definitions.
- Call out uncertainty instead of smoothing over weak evidence.
- If evidence is weak, say what transcript segment or timestamp needs review instead of guessing.
- Finish with a concise artifact I could paste into my learning app.
Misconceptions
What to stop believing.
Creative AI removes the need for taste.
It increases the need for taste because output volume explodes.
The best prompt is enough.
References, critique, iteration, and post-production matter just as much.
Practice studio
Learning only counts when you make something.
01
Transcript evidence map
Separate what the video actually says from what you already believe about the topic.
3 source-backed takeaways with timestamps, confidence, and a transfer note.02
One useful artifact
Apply the video to a real workflow and produce a creative workflow board with critique criteria and review checkpoints..
A reusable artifact with a done signal and one verification step.03
Teach-back card
Explain the lesson to someone who has not watched the video yet.
A 90-second explanation, one diagram, one example, and one misconception to avoid.
Recall check
Answer first, then reveal — without rewatching.
How did LongCat 2.0 gain popularity before its official launch?
What is LongCat 2.0's biggest weakness and how should you use it instead?
How does the architect-executor pattern save money?
Source shelf
Use the video as a doorway, then verify with primary sources.
