Alibaba prices Qwen3.8-Max at $2 per 1M input tokens and $6 per 1M output tokens via its API, below Kimi K3's $3/1M input tokens and $15/1M output tokens
Chinese tech giant Alibaba Group made its new flagship model, Qwen3.8-Max, widely available on Monday through application programming interfaces …
Context & Ripple Effects
This is the commercial follow-through to Alibaba’s Qwen3.8-Max preview, which positioned the 2.4T-parameter model against frontier systems and foreshadowed open-weight availability. API access turns that positioning into a directly comparable operating cost for developers.
Alibaba has previously cut Qwen API prices amid China’s AI price competition. The new flagship’s published rates now put a concrete price gap beside Kimi K3, while Alibaba has separately said Qwen3.8-Max performs better on some benchmarks.
First-order effects
- Developers can deploy Qwen3.8-Max through an API at the stated $2-per-million input-token and $6-per-million output-token rates; output-heavy workloads face a particularly large listed-price gap versus Kimi K3.
- Kimi must now defend a higher published price point with performance, reliability, ecosystem, or other workload-specific advantages; the reported pricing alone favors Alibaba on token spend.
Second-order effects
- Chinese model providers competing for API workloads face greater pressure to match prices, offer discounts, or differentiate their models beyond headline benchmark claims.
- Application builders that generate substantial model output have more room to test premium-capability models in production, because output-token pricing has a larger effect on their recurring inference bill.
Third-order effects
- If frontier-class models continue to reach APIs at sharply differentiated token prices, model selection will increasingly turn on effective cost per useful task rather than benchmark rank or parameter count alone.
- Sustained price competition could shift value toward providers with the capacity and distribution to absorb lower inference margins, while forcing smaller API vendors to specialize or find other revenue sources.
The trend: This is another data point in the shift from frontier-model launches as product events to inference economics as a central competitive lever.