# elvis: StepFun Step 5 Preview matched GLM 5.3 on two real coding tasks

elvis (@omarsar0) says he early-tested StepFun’s new Step 5 Preview as a coding agent. He reports that it and GLM 5.3 both solved two real repository tasks with no fix rounds, that Step 5 Preview stopped after checking its work, and that it found five hidden clues in a ~368K-token long-context test.

Language: en
Time zone: America/Los_Angeles
Current time (UTC, cached as of): 2026-11-10T07:31:00.000Z
Query date (America/Los_Angeles): 2026-11-09

HTML: https://didcodexreset.com/news/f816f9c30fac41ba87d55a63.html

Content language: en
Localization state: sameLanguage

Source: elvis · Published 9/21/2026, 07:08:22

elvis (@omarsar0) says he early-tested StepFun’s new **Step 5 Preview** model as a coding agent. He describes it as comparable to GLM 5.3, Kimi K3, and others, at a competitive price, and says he tested it in a minimal harness against **GLM 5.3**. He also calls it on the “Pareto frontier for cost vs. capability.” The post thanks the StepFun team for partnering on it.

He reports giving both models two real tasks in the same repo at the same commit: a bug where numeric filters silently returned zero rows for decimals and negatives, then a feature that needed new routes, permission gating, and a refactor of the background task supervisor. According to him, both models got both tasks right, every held-out test passed with no regressions, and neither received a follow-up prompt or retry. **Step 5 Preview** finished, checked its work, and declared itself done both times, while **GLM 5.3** wrote correct code and then kept going until the step limit ended the run. On the bug, he says Step 5 Preview wrote the shorter patch, matching the approach in the real Datasette maintainer commit, and added its own tests without being asked.

In a separate long-context test, he says he generated about **368K tokens** of fake incident tickets and hid five clues that together explain an outage; when asked for the root cause, Step 5 Preview found all five clues and connected them correctly in about 90 seconds. Based on this, he says he would reach for it on unattended agent runs where a clear completion signal matters more than speed, on bug fixes in unfamiliar codebases, and on long-context work.

Tags: StepFun, Step 5 Preview, Coding agents, Model eval

[View original post](https://x.com/omarsar0/status/2102035017262149640)