DEVOURED
StepFun's Step 5 Preview
StepFun’s Step 5 Preview achieves top-tier reasoning performance while operating at significantly lower costs than rivals like Kimi K3.
What: Step 5 Preview is a 600B parameter MoE model that scores 44 on the Artificial Analysis Intelligence Index. Despite its competitive reasoning performance, it lags behind leaders like Qwen3.8 Max and GLM-5.3 in agentic tasks like terminal interaction.
Why it matters: The model demonstrates that pricing and token efficiency are becoming as significant as raw intelligence for enterprise adoption, with StepFun significantly undercutting major competitors.
Decoder
- MoE (Mixture of Experts): A model architecture where only a subset of parameters ('experts') is activated for each token, improving efficiency.
- Pareto Frontier: In this context, the set of models that offer the best possible intelligence score for a given cost level.
Original article
Full article content is not available for inline reading.
Explore all 44 articles in today's digest
Fresh dev news, AI-digested daily. Never miss a trend at devoured.fyi