⚙️ AI Hardware

StepFun's 196B Beast Runs Top AI Scores for Pennies—And It's Open Source

Imagine running AI that smokes the leaders without bankrupting your GPU budget. StepFun's Step-3.5-Flash just made elite performance dirt cheap for devs everywhere.

Sarah Chen 📅 Mar 24, 2026 ⏱️ 3 min read 👁️ 8 views
Bar chart of Step-3.5-Flash crushing Kimi K2.5 and others on decoding cost and benchmarks

⚡ Key Takeaways

  • Step-3.5-Flash tops open-source on math, code, agents at 1/19th rivals' inference cost.
  • 11B active params from 196B total via smart MoE—efficiency breakthrough.
  • China's quiet labs outpacing West's hype machines; open-source wins for real users.

🧠 What's your take on this?

Cast your vote and see what theAIcatchup readers think

Sarah Chen
Written by

Sarah Chen

AI research editor covering LLMs, benchmarks, and the race between frontier labs. Previously at MIT CSAIL.

#AI benchmarks #MoE models #Step-3.5-Flash #open-source AI

Worth sharing?

Get the best AI stories of the week in your inbox — no noise, no spam.

Originally reported by Towards AI

Related Stories

📬

Stay in the loop

The week's most important stories from theAIcatchup, delivered once a week.

No spam. Unsubscribe any time.