StepFun joins the frontier: a previously non-frontier Chinese lab (StepFun) released a Kimi K3-level model, 3 times cheaper per Artificial Analysis
Summary
StepFun, a Chinese AI lab, has released Step 5 Preview, which matches the performance of Kimi K3 at one-third the cost, marking its entry into the AI frontier.
Similar Articles
@heyshrutimishra: China undercut the entire Western AI pricing model. Kimi K3 matches Claude Fable 5 on coding benchmarks, but output tok…
Chinese AI model Kimi K3 matches Claude Fable 5 on coding benchmarks but costs a third of the price, signaling a structural collapse in the cost of intelligence. Moonshot plans to release open weights on July 27, further pressuring Western pricing models.
Less Than a Month: Kimi K3, Qwen3.8, DeepSeek-V4-Pro-0813, GLM-5.3
Multiple leading Chinese AI labs have released new flagship models within the past month, including Kimi K3, Qwen3.8, DeepSeek-V4-Pro, and GLM-5.3, signaling rapid progress in China's AI race.
Stepfun new model "Step 5 Preview" just leaked
Stepfun's new AI model, Step 5 Preview, has been leaked, showing a significant version jump from 3.7 to 5.0 with major performance improvements.
Kimi K3: Open Frontier Intelligence
Kimi introduces Kimi K3, a 2.8 trillion parameter open model with native vision and 1M context, built on Kimi Delta Attention and Attention Residuals. It achieves frontier-level performance in coding and reasoning, with full weights to be released by July 2026.
AI usage is getting expensive and cheaper as well. The 2026 Frontier Showdown
Two frontier AI models, Moonshot AI's Kimi K3 (2.8T open-weight MoE) and Alibaba Cloud's Qwen 3.8 Max (2.4T sparse MoE), were released in July, offering different strengths: Kimi K3 excels in speed and cost control while Qwen 3.8 Max provides autonomy and multimodality.