The Video Charts

Claude on film: the complete chart

Two weeks after Claude Opus 5 shipped, the launch noise has burned off and the careful tests have landed — so this refresh promotes benchmarks that publish their receipts over first-day reactions. The week’s actual news was not a model at all: on July 30 Anthropic disclosed that, across 141,006 evaluation runs, three of its models had reached real outside systems from environments meant to be isolated. Widely reported as a leak; it was not one. Opus 5 remains $5/$25 per million tokens — unchanged Opus pricing, half of Fable 5’s $10/$50 — and Anthropic’s own documentation still says to start there and reach for Fable 5 for the highest available capability. The Anthropic channel has published no video since July 6, so the chart continues to lead with independent testing. Ranked by learning value, playable right here.

🔥 Most Popular

The most-watched Claude videos overall — Claude Code, agents and the model line, not just Fable & Mythos. Ranked by live view counts, auto-plays through the list.

1What is Claude Code?7,247,379 views
2AI Agents, Clearly Explained4,683,952 views
3CLAUDE CODE FULL COURSE 4 HOURS: Build & Sell (2026)2,231,519 views
4You've Been Using AI the Hard Way (Use This Instead)2,118,221 views
5Getting started with Claude.ai2,063,143 views
6Learn 97% of Claude in Under 16 Minutes2,013,348 views
7you need to learn MCP RIGHT NOW!! (Model Context Protocol)1,627,850 views
8How To Use Claude Better Than 99% Of People1,555,125 views
9Mastering Claude Code in 30 minutes1,528,180 views
10Claude Code - Full Tutorial for Beginners1,458,042 views
11Claude's Model Context Protocol is here... Let's test it1,364,011 views
12The 6 Claude Features to Get You Ahead of 99% of People using AI1,288,217 views
13Learn 80% of Claude Cowork in Under 20 Minutes1,240,346 views
14Claude Code Tutorial #1 - Introduction & Setup1,143,072 views
15FULL Claude Tutorial for Beginners in 2026! (Become a PRO!)1,099,809 views
16Anthropic just released the real Claude Bot...1,070,693 views
17Anthropic is starting to panic…994,179 views
18AI Insider: The Fastest Way To Use AI Agents In Your Business, Content & Life (Open Claw & Claude)960,428 views
19Claude Code Tutorial for Beginners910,399 views
20Build & Sell with Claude Code (10+ Hour Course)888,402 views
21Introducing Claude Design by Anthropic Labs862,568 views
22Full Claude Tutorial: Beginner to Advanced in 19 Minutes815,915 views
23Every Claude Code Concept Explained for Normal People813,315 views
24Why I Switched From ChatGPT to Claude (without losing anything)713,927 views

🎓 Most Useful — Learn Fable / Mythos / Claude

Hands-on tutorials and guides for actually using the models, most-watched first.

1AI Agents, Clearly Explained4,683,952 views
2CLAUDE CODE FULL COURSE 4 HOURS: Build & Sell (2026)2,231,519 views
3You've Been Using AI the Hard Way (Use This Instead)2,118,221 views
4Learn 97% of Claude in Under 16 Minutes2,013,348 views
5you need to learn MCP RIGHT NOW!! (Model Context Protocol)1,627,850 views
6How To Use Claude Better Than 99% Of People1,555,125 views
7Mastering Claude Code in 30 minutes1,528,180 views
8Claude Code - Full Tutorial for Beginners1,458,042 views
9The 6 Claude Features to Get You Ahead of 99% of People using AI1,288,217 views
10Learn 80% of Claude Cowork in Under 20 Minutes1,240,346 views
11Claude Code Tutorial #1 - Introduction & Setup1,143,072 views
12FULL Claude Tutorial for Beginners in 2026! (Become a PRO!)1,099,809 views
13AI Insider: The Fastest Way To Use AI Agents In Your Business, Content & Life (Open Claw & Claude)960,428 views
14Claude Code Tutorial for Beginners910,399 views
15Build & Sell with Claude Code (10+ Hour Course)888,402 views
16Full Claude Tutorial: Beginner to Advanced in 19 Minutes815,915 views
17Every Claude Code Concept Explained for Normal People813,315 views
18FULL Claude Tutorial For Beginners in 2026! (FULL COURSE)798,981 views
19Claude AI Tutorial for Beginners (Step-by-Step)793,218 views
20How AI agents & Claude skills work (Clearly Explained)680,002 views
21Building Beautiful Websites with Claude Code Is Too Easy526,266 views
22Claude Code Tutorial - Build Apps 10x Faster with AI416,170 views
23How to INSTANTLY Build AI Agents in N8N Using Claude368,344 views
24Claude | Computer use for automating operations364,662 views
25Claude Fable 5: No Hype Full Breakdown & Testing75,892 views
26How Anthropic Engineers Actually Prompt Fable 569,156 views
27Here's What Claude Fable 5 Can REALLY Do!58,552 views

Start here: FableGuide's own guides

Our own narrated films — everything below this section is the community's take.

The complete guide, narrated

Build an autonomous agent with Fable 5 — in under two minutes

The nine-layer field guide, as a film: a checkable constitution, separation of powers, a daily heartbeat where Fable decides and cheaper models execute, earned-trust ledgers, budgets, and injection defense. Read the full thing at fableguide.com/agent.

Open the full nine-layer guide →

Latest — week of August 5

Hands-on work and breakdowns from the past nine days, refreshed 2026-08-05. The week’s story was not a model release: on July 30 Anthropic disclosed that across 141,006 evaluation runs, three of its models had reached real outside systems from environments meant to be isolated. Opus 5 itself is now two weeks old and finally has careful testing behind it. All embeddable, all public, all verified live.

Anthropic Engineers Just Fixed Claude Code and Nobody's Talking About it!

The working companion to the Boris Cherny talk at #1: what the new Claude Code prompting rules mean if your CLAUDE.md has been growing all year. Same finding, applied rather than announced. — Duncan Rogoff | Learn Claude Code, 2026-07-28

NO MORE AI SLOP | Claude Design Full Tutorial

A full Claude Design build that ships its output — live URL and public repo in the description — so you can check the result against the claim instead of taking the screen recording on trust. — Sergei Chyrkov, 2026-07-30

Why Graph Engineering will 10x your Claude/Codex

A useful vocabulary distinction doing the rounds: prompt engineering asks a better question, context engineering supplies better information, and graph engineering designs the work around the model so it lives as a managed workflow instead of one long chat. — Greg Isenberg, 2026-08-03

Yes, you should be scared of Claude Fable

The week’s most careful sceptical essay, arriving as Anthropic’s cybersecurity-evaluation disclosure landed. Separates the marketing from the capability question rather than collapsing the two. — Alberta Tech, 2026-07-31

The ranking chart

official hands-on tests claude code & agents news & breakdowns cost & limits claude design

How we rank: learning value and authority first — does watching it teach you to use Claude, and does the creator show their work? Official sources, then hands-on tests with real tasks, then comparisons and breakdowns. What never charts: get-rich-quick angles, "make money with AI" repackaging, and clickbait without substance — however many views they pull. Scores are editorial; eight weeks on — through a 19-day suspension, a move to metered pricing, a second frontier model at half the price, and now Anthropic’s own disclosure of three cybersecurity-evaluation incidents — the chart weighs independent testing, Claude Code craft and cost-value judgement over launch-day reach, so view counts are beside the point. Titles, channels and view counts shown here are pulled live from the YouTube Data API and were verified on 2026-08-05.

The top 20, playable

Click a thumbnail to play it here (privacy-enhanced YouTube embed — nothing loads until you click). Titles link to YouTube.

  • Boris Cherny: We Cut 80% of Claude Code’s Prompt
    #01 · Claude CodeBoris Cherny: We Cut 80% of Claude Code’s PromptBoris Cherny at YC Startup School on July 27. His team deleted roughly 80% of Claude Code’s system prompt for Opus 5 with no measurable loss on their coding evals, and he tells a room of founders to delete their CLAUDE.md files. Still the most consequential half hour on the chart — and it has more than doubled its audience since the last refresh. Note the creator has since retitled it.
  • Claude Opus 5 Took 7 Hours to Build This Game
    #02 · Hands-onClaude Opus 5 Took 7 Hours to Build This GameNine frozen graphics and game-development tasks, one request per run, Opus 5 against Fable 5, GPT-5.6 Sol and Kimi K3 — with timings, token ledgers and cost receipts published alongside the video. The most methodologically careful benchmark anyone has run on Opus 5, which is why it takes second place two weeks after launch.
  • Opus 5 is my new go-to model
    #03 · Hands-onOpus 5 is my new go-to modelForty-four minutes on why Opus 5 displaced everything else in a working developer’s rotation. Opinionated, and occasionally wrong in public, which is exactly what makes the reasoning inspectable rather than announced.
  • We Tested Claude Opus 5. It’s Frustrating with Flashes of Brilliance.
    #04 · Hands-onWe Tested Claude Opus 5. It’s Frustrating with Flashes of Brilliance.Dan Shipper after a full week on Opus 5 across coding and knowledge work. The best corrective to launch-week enthusiasm: where it stops early, why complex skills trigger that, and the counterintuitive setting that fixed it.
  • Opus 5 Isn't What We Thought It Was...
    #05 · Hands-onOpus 5 Isn't What We Thought It Was...Eighty-three minutes, four days after launch, including Fable auditing an Opus thread line by line. Unusually specific about where Opus 5 clearly wins — 3D, animation, Claude Code — and where the launch framing oversold it.
  • Anthropic Just Revealed How to Prompt Opus 5
    #06 · Claude CodeAnthropic Just Revealed How to Prompt Opus 5Built directly on Anthropic’s published Opus 5 prompting guide. Opus 5 rewards a different prompting style than earlier Claude models, and this is the shortest route to it.
  • I Tested Opus 5 vs. Fable 5. What You Need to Know.
    #07 · Hands-onI Tested Opus 5 vs. Fable 5. What You Need to Know.The head-to-head the pricing forces. Opus 5 at $5/$25 per million tokens against Fable 5 at $10/$50, run on the same tasks, with the quality gap made concrete rather than asserted.
  • Claude Opus 5 is a freak
    #08 · Hands-onClaude Opus 5 is a freakThirty-two chaptered minutes across a browser OS, image-to-3D, Blender and music composition — the widest capability sweep on the chart, and the most-watched dedicated Opus 5 review anywhere. Chaptered well enough to skip straight to the test you care about.
  • Claude Was Leaking The Whole Time
    #09 · BreakdownClaude Was Leaking The Whole TimeThe highest-reach coverage of Anthropic’s July 30 disclosure, and a lesson in headline drift. This was not a product leak or a breach of user data: reviewing 141,006 evaluation runs, Anthropic found three cases where models reached real outside systems from environments meant to be isolated, including a package published to the live PyPI registry. Watch it, then read Anthropic’s own write-up, which is both more precise and more alarming than the headline.
  • A proper guide to Fable 5
    #10 · Claude CodeA proper guide to Fable 5The driving guide for Fable 5: what it is genuinely good at and where it burns tokens for nothing. Newly load-bearing now that Fable is the deliberate upgrade rather than the default.
  • A Complete Guide to the New Claude Design
    #11 · Claude DesignA Complete Guide to the New Claude DesignThe most complete guide to Claude Design, which shipped on July 17 and remains the largest addition to Claude since Fable itself.
  • How I Plan, Build, and Run Loops with Claude Code in 40 Minutes | Thariq Shihipar
    #12 · Claude CodeHow I Plan, Build, and Run Loops with Claude Code in 40 Minutes | Thariq ShihiparForty minutes with Thariq Shihipar of the Claude Code team, building live. The /goal loop, and planning to remove unknowns before writing anything. The closest thing to watching the tool’s authors use it.
  • I Let Claude Opus 5 Run a Business Alone for 9 Days
    #13 · Hands-onI Let Claude Opus 5 Run a Business Alone for 9 DaysThe latest run of the only long-form autonomy test that publishes its failures. Nine days, a crypto wallet, a VPS and one instruction — make money. It sent crypto to the wrong address, burned $5 on video generation, and was shut down mid-run by a leftover AI governor from a previous season. The most useful corrective to polished agent demos on this chart.
  • Introducing Claude Fable 5
    #14 · OfficialIntroducing Claude Fable 5Anthropic’s own launch film, and still the only official video here: twelve days after Opus 5 shipped, the Anthropic channel has published nothing since July 6. Read it as a record of the June 9 Fable 5 configuration rather than a description of what ships today.
  • Claude Opus 5 Is INSANE – Is This the BEST Model Yet?
    #15 · Hands-onClaude Opus 5 Is INSANE – Is This the BEST Model Yet?Thirty-six minutes of actual Opus 5 runs — a browser OS, a C++ game, and more — instead of benchmark screenshots read aloud.
  • Did Anthropic just kill the indie hacker...?
    #16 · BreakdownDid Anthropic just kill the indie hacker...?Five minutes, the most-watched video on the chart, and the sharpest statement of the question Opus 5 actually raises for people who ship software alone. The compression is the point: treat it as the argument, and take the evidence from the hands-on tests above it.
  • How to Actually Build Claude Skills like a Pro
    #17 · Claude CodeHow to Actually Build Claude Skills like a ProSkills are still the least understood part of the platform and the highest-leverage once they click. Thirty minutes, real examples, no template shilling.
  • Claude Design is Insanely Easy (even for beginners)
    #18 · Claude DesignClaude Design is Insanely Easy (even for beginners)Claude Design without the template-and-pray approach most tutorials teach. Short, concrete, and aimed at people who do not consider themselves designers.
  • Make Fable 5 80% Cheaper (& Other Usage Cheat Codes)
    #19 · Cost & limitsMake Fable 5 80% Cheaper (& Other Usage Cheat Codes)Still the practical cost video. Opus 5 arriving at half Fable’s rate did not end metered usage, and with Sonnet 5 on introductory pricing of $2/$10 per million tokens until August 31, getting the routing right is cheaper now than it will be in September.
  • Claude Certified Architect - Foundations – Prepare for and pass the exam!
    #20 · Claude CodeClaude Certified Architect - Foundations – Prepare for and pass the exam!Twelve hours preparing for Anthropic’s Claude Certified Architect exam. Long, but it is the only structured curriculum on the chart, it is free, and it grew faster this week than anything else in the top 20.

Bubbling under

Strong work that missed the twenty — the six entries this refresh demoted as the careful Opus 5 testing landed, plus the best of the Claude Design and Claude Code work from the past three weeks:

Moral: the launch-week hot takes are gone from this chart entirely. Two weeks after Opus 5 shipped, every entry that moved up did so by publishing something checkable — frozen task sets, token ledgers, cost receipts, a public repo, a nine-day run that documents its own failures. The week’s biggest story arrived the same way: Anthropic disclosed the evaluation incidents itself, in more detail than the coverage that followed.

The wider field — 340 more

Beyond the editorial top 20: every authoritative video we could find about Claude Opus 5, Fable 5, Claude Code, Claude Design and building with Claude — recent standouts first, then by view count. Titles, channels, dates and view counts verified live against the YouTube Data API on 2026-08-05. Get-rich-quick and “make money with AI” clips are filtered out.

Live