TLDRocket
Sign in

Step 5 Preview

Product Hunt Zac Zuo ● Covered by 2 sources

StepFun launched Step 5 Preview, an open-weight AI model for coding, finance research and long-running agents. It’s built for huge context windows and runs with common deployment tools.

Based on reporting by Product Hunt, Zac Zuo — read the original for the full story.

Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error

StepFun is back with its third launch, and this one is built for people who want an open model that can stay on task for a long time. Step 5 Preview is an open-weight mixture-of-experts model aimed at coding, financial research, and long-running agents.

The headline numbers are big: 600B total parameters, 27B active per token, and a 1M-token context. That last bit is the real hook. It points to a model meant to hold a lot more in memory than the average chatbot prompt-and-pray setup.

StepFun is also making the model easier to plug into existing stacks. It has deployment support for vLLM, SGLang, and Transformers, which should make it more familiar territory for teams already working in those tools.

The launch framing is pretty clear. This is not a general “look what we made” demo. It’s an open-weight system pitched for specific, messy jobs where models have to keep going without losing the thread.

My take — AI-written commentary, not fact-checked reporting

Open-weight models keep winning the practical argument: if a system is meant for real workflows, people want to inspect it, run it, and move on. The closed-model crowd can keep the glossy demos; the rest of the industry still has actual work to do.

Read more about this at: Product Hunt

Related stories

The daily briefing

Every AI story that matters, in your inbox by 8am.

TLDRocket reads all relevant sources, removes duplicate coverage, and summarises the day in two minutes. Follow companies and topics for alerts, or get the briefing in Slack. Free, no spam, unsubscribe anytime.