Stepfun just announced their new SOTA model
This is a dev post classified by Jev as AI dev tools (a launch), kept by the Dev Radar because it carries real work, not commentary.
Stepfun just announced their new SOTA model step 5 preview is a 600b model with only 27b active params, and early numbers are looking insane: > 67.7 on deepswe > beats gemini 3.8 flash > 1m context window + image and video input > $1 input and $2.70 output > beats most of chinese models it’s already available in api and chat, so you can play around if you want already got access and running some tests now, will post results soon which model should i test it against?
Posted by J A Z I I (9.7k followers) 1 h ago · 18 likes · 727 views · view the original post on X. Kept by the Dev Radar as AI dev tools. Tools mentioned: StepFun Open Platform.
More dev work like this
- Your coding agents shouldn’t start every session from zero. — @DanKornas
- "SoL-Pi: Recursively Scaling Auto-Research Loops for Efficient Agent Harness" — @askalphaxiv
- A 27B AI model running locally inside a browser.🚀 — @BkashJosi
- Introducing a new way FIND (⌘F) with @typesafeai Jev. — @Saboo_Shubham_
- Codex 在家里的 Mac 上跑着一个长任务，人出门了，跑到要审批的那一步就停在那儿等，回来才发现半天没动。 — @GitHub_Daily
- Coding agents get expensive when every task gets premium-model treatment. — @DanKornas
- iMessage was never meant to win. — @xadisingh
- Someone squeezed pretraining, SFT, reward modeling, DPO, PPO and GRPO into a… — @agenticgirl
Every post is read and classified by Jev (TypeSafe): what it is, which market it belongs to, and whether the link is a real tool. 15.3k posts from 4.8k X accounts over the last 21 days, 1.7k tools, 12 markets. Collected every 5 minutes, fully re-ranked every hour — last update 2026-09-20 08:11 UTC. Full method.