About The Feature Crew

---

We're tech professionals working on AI products. Our goal with this channel is to provide a grounded but accessible take on AI -- and have a few laughs along the way.

⚠️ Important
This channel is in no way affiliated with Microsoft, or any of our previous employers. Our opinions are our own and not professional advice.

Claude 4.5 sonnet surprised us...

Today we're testing Claude 4.5 sonnet - Anthropic's new workhorse agentic model touting impressive performance on long-running tasks. Chapters 0:00 Intro 0:22 Coding 8:15 Business 19:02 Maze 24:55 Closing Thoughts

---

Does GPT-5 Codex live up to the hype?

Today we are testing GPT-5 codex to see if it is really the best coding agent on the market. ANNOUNCEMENT If you like these tests, check out Jakob's new creativity-focused benchmark at https://vibebench.ai PROMPTS Planet https://featurecrew.io/prompts/planet/v1.txt https://featurecrew.io/prompts…

---

Is GPT 5 PRO worth $200? Let's find out...

Today we're putting gpt-5 pro to the test to see if it is really worth the $200/mo pricetag of ChatGPT pro. Chapters 0:00 Intro 0:32 Coding 8:43 Business Reasoning 16:07 Conclusions and Bonus Test

---