Why it worked
The video leverages the excitement around a new AI model release (Claude Opus 4.8 and Mythos) and highlights key features like "dynamic workflows" and improved performance, and potential for complex task automation, which appeals to users interested in cutting-edge AI technology.
Summary
The video announces the release of Opus 4.8 and introduces "dynamic workflows" for Claude, highlighting its increased autonomy for complex tasks. It also teases the upcoming "Mythos" model, noting its powerful capabilities and potential for high token usage.
Structure
- 1Announcement of Mythos and Claude's release
- 2Introduction of Opus 4.8 and benchmarks
- 3Explanation of dynamic workflows
- 4Discussion of Mythos's capabilities and token usage
- 5Mention of safeguards for Mythos
Product placement
The video mentions and shows benchmarks for "Claude Opus 4.8" and "Claude Mythos Preview". It also discusses "dynamic workflows" in the context of Claude.
On-screen text
Opus 4.8
Opus 4.8
Opus 4.7
GPT-5.5
Gemini 3.5 Pro
69.2%
64.2%
58.6%
54.2%
Agentic coding
SMS Responser
Agentic terminal coding
Terminal Bench 2.1
74.6%
45.9%
78.2%
70.3%
Multi-disciplinary reasoning
Probability/Exact Match
49.8%
45.9%
41.4%
44.4%
82.9%
54.7%
52.2%
51.4%
Agentic computer vision
Unsupervised VQA
83.4%
82.8%
78.7%
76.2%
Knowledge work
GOPS AA
1890
1793
1368
1314
Agentic financial analysis
Finance Agent v3
53.9%
51.5%
51.8%
43.0%
A
Claude Opus 4.7 vs 4.8 side-by-side canvas test
Claude Opus 4.7 vs 4.8
Claude Opus 4.7
Claude Opus 4.8
10:47 AM - May 28, 2026 - 97.7K Views
My colleague's "dynamic workflows" are, in my opinion, the most significant
Claude Code innovation in 2026 so far. Unpacking "Claude dynamically
writes orchestration scripts" from the blog post: Claude has the flexibility to
determine "phase" of subagents to "run" and "prompt" for "agents"
leveraging your session's context "and" Claude's existing ability to write
scripts. Workflows massively reduce the prompting you have to do to
launch complex agent processing.
For reviewing code, I like to set up a workflow with a phase that
brainstorms potential issues, a second phase to research each issue in
detail, and a third to resolve the issues that are verified through the
research. This process gives me much more confidence in my code quality.
That said, it will use up lots of tokens.
What's next?
Users will find Opus 4.8 to be a modest but tangible improvement on its
predecessor. There's still more to be done: we're working on developing and
releasing models that provide many of the same capabilities as Opus at a lower
cost.
Not only that, but we plan to release a new class of model with even higher
intelligence than Opus. As part of Project Glasswing, a small number of
organizations are currently using Claude Mythos Preview for cybersecurity
work. Models of this capability level require stronger cyber safeguards before
they can be generally released. We're making swift progress on developing these
safeguards and expect to be able to bring Mythos-class models to all our
customers in the coming weeks.