The most interesting thing about this release, to me, is the implication that there had…
This is a AI post classified by Jev as Frontier models (a model release), kept by the AI Radar because it carries real work, not commentary.
The most interesting thing about this release, to me, is the implication that there had been Step 4 which never made it. it's plausible that it's a ≈GLM 5.3 level model. If the architecture is okay, they could do RL faster than Kimi/ZAI/Xiaomi. They need to shed Preview though.
Posted by Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞) (76.3k followers) 1 h ago · 33 likes · 2.4k views · view the original post on X. Kept by the AI Radar as Frontier models. Tools mentioned: StepFun Open Platform, 阶跃星辰.
More AI work like this
- Step 5 Preview from @StepFun_ai is 600B params total where only 27B are active, and… — @plotarmordev
- DeepSeek 2.0 Leak: Coming Soon 🔥 — @Priyannkaaaa
- Introducing Da7em Bench. — @Da7_Tech
- Claude Fable 5.2 @claudeai @ClaudeDevs you guys seriously cooked with this model — @chetaslua
- Chat GPT-6 Astra solve 108 year old unsolved WWI german code for first time in history. — @Vvikramai
- We’ve been quiet for too long. It’s time for a change. — @sheriyuo
- Introducing Step 5 Preview: Advancing the Pareto Frontier. — @StepFun_ai
- Ran FlappyBench on Qwen3.8-Omni-Flash, DeepSeek-V4.1-Flash, and Gemini 3.8 Flash with… — @CommandCodeAI
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. 31.8k posts from 4.8k X accounts over the last 14 days, 1.4k tools, 19 markets. Collected every 5 minutes, fully re-ranked every hour — last update 2026-09-20 05:05 UTC. Full method.