AI news · October 10, 2026
StepFun's Step 5 Preview offers 1M context for $1 input per million
Step 5 Preview claims a 1M-token context at $1/M input and $2.70/M output, with OpenRouter access; builders can compare long-context coding against Claude Opus 5 on a real task.
- context tokens
- 1M
- input tokens
- $1/M
- output tokens
- $2.70/M
TLDR AI's Oct 8 issue highlighted StepFun's Step 5 Preview as a sponsored launch. The listed price is $1 per million input tokens and $2.70 per million output tokens, with a one-million-token context. StepFun positions it for coding agents, financial analysis, and long-running work, and says the model is available through OpenRouter, OpenCode, Kilo Code, Cline, omp, and Hermes Agent.
The sponsor note says Step 5 Preview scores near Claude Opus 5 on Artificial Analysis's Intelligence Index, but that is a launch claim rather than a neutral comparison with Claude Opus 5.5 or Microsoft's new Decision-1. Large context can reduce document splitting and repeated retrieval, but it does not guarantee useful reasoning. Test the same long task on your current model before changing a production route.
What you can do with it
Try Step 5 Preview through OpenRouter or one of the listed coding integrations. Run the same two or three long-context coding or document tasks on your current model, then compare quality, latency, and total cost; the listed rates are $1/M input and $2.70/M output.
Our take
The price is the story: a long-context model at this rate can change the cost of batch work. The catch is that the near-Opus claim comes from a sponsored launch note and a named index, not a neutral benchmark, so only your own tasks can justify a switch.
Links TLDR launch noteTry on Vercel AI Gateway
Source: TLDR AI — Made With Models writes the brief; the reporting is theirs.