Navigate Select ESC Close

Claude Opus 4.8: Here is Everything that Changed

2026-05-28 Science & Technology
11.3k
237
34
Prompt Engineering
Prompt Engineering
241.0k subscribers

Unlock all features

FREE: Get instant access to 10 AI summaries, chats, or transcripts per day.

Description

Checkout the AI Learning paths: https://jb.gg/academy/aws-ai-paths In this video I break down Anthropic’s Claude Opus 4.8 release, which they frame as an incremental upgrade over 4.7 but optimized for long-running tasks. I walk through the key new features: dynamic workflows that can spawn hundreds of parallel sub-agents for verifiable work like large code migrations, the return of manual effort control (low to max) replacing adaptive thinking, and a Messages API update that lets developers modify system instructions mid-task without breaking prompt cache. I also cover benchmark highlights and why the harness used can change results, discuss pricing (still $5/M input and $25/M output) plus a big Fast Mode price drop (2.5x speed, now 3x cheaper), and run quick claude.ai tests at different effort levels. I close with thoughts on faster release cycles and a hint about Mythos reaching more customers soon. LINKS: https://www.anthropic.com/news/claude-opus-4-8 Prompt Caching Video: https://youtu.be/HDMqDV7mmGo Harnes Engineering: https://youtu.be/uY9tMU-KS4A https://www.youtube.com/@JetBrainsAcademy My voice to text App: whryte.com Website: https://engineerprompt.ai/ RAG Beyond Basics Course: https://prompt-s-site.thinkific.com/courses/rag Signup for Newsletter, localgpt: https://tally.so/r/3y9bb0 Let's Connect: 🦾 Discord: https://discord.com/invite/t4eYQRUcXB ☕ Buy me a Coffee: https://ko-fi.com/promptengineering |🔴 Patreon: https://www.patreon.com/PromptEngineering 💼Consulting: https://calendly.com/engineerprompt/consulting-call 📧 Business Contact: [email protected] Become Member: http://tinyurl.com/y5h28s6h 💻 Pre-configured localGPT VM: https://bit.ly/localGPT (use Code: PromptEngineering for 50% off). Signup for Newsletter, localgpt: https://tally.so/r/3y9bb0 00:00 Opus 4.8 Overview 01:15 Benchmarks and Behavior 02:36 New Features Breakdown 04:28 Mythos and Pricing 06:12 Sponsor Segment 08:06 Benchmark Harness Matters 09:11 Dynamic Workflows Demo 11:40 Fast Mode Price Drop 12:06 Quick Model Tests #anthropic #jetbrains #sponsored #claude

Top Comments (10)

@duacot6633 2026-05-28

Installed Claude desktop on my PC and thought i picked up malware. Four days later i find that it was the desktop app for Claude that added an invalid registry line that was causing the problem. Seems odd that this happened.

0
@Robinho195 2026-05-28

which terminal are you using for the CLI? Is it Mac only?

0 1 replies
@Aubfred89 2026-05-28

Us poor folk still waiting to hear about the next Sonnet Model

12 6 replies
@longboardfella5306 2026-05-28

Thanks. Btw - Adaptive not adoptive. Totally different word and meaning. Just to help

7 1 replies
@aiwold_solutions 2026-05-28

Always better then previous only competitors models how.

0
@MrAlroyNdhlovu 2026-05-28

It seems to have a 200k context window and not a 1m. Maybe its just me.

0 2 replies
@trappedcat3615 2026-05-28

I guess we get a new release monthly now. Micro improvements. Gemini is the only one that waits a season.

0
@beliebigerusername 2026-05-28

"Hello." => 5-hour limit reached

3 1 replies
@OutdoorswithLittleBuddy 2026-05-28

i bet gpt 5.6 will be better then this or mythos preview!

4 1 replies
@metrodyne 2026-05-29

Everything is about code and terminal nowadays, but what about health knowledge, writing and understanding big contexts without context rotting earlier? What about psychology and business expertise? Damn...

0

Unlock the Data Inside
Turn Videos into Knowledge

  • Get FREE 10/day: transcripts, summaries, chats
  • Chat with videos, export text & PDF
  • $1 free API credit for RAG, chatbots & research

Free forever plan • All features unlocked

App screenshot