AIReiter flags a meaningful API-contract gap for Qwen3.8-Max-0902: while Alibaba's September 2 X announcement promotes the upgrade as live through QwenCloud, the Alibaba Cloud Model Studio page was updated the same day but still labels the model as "qwen3.8-max" with no separate qwen3.8-max-0902 SKU, pricing row, or be The documented envelope on the public qwen3.8-max page is 2.4-trillion-parameter MoE, 1M-token context, multimodal input with text output, function calling, structured outputs, partial mode, and context caching across the listed regions, with fine-tuning unsupported everywhere. AIReiter notes that no matched -0902 benc

Command Code's pricing page for Qwen 3.8 Max 0902 confirms the same $2/$6 per 1M input/output pricing as the base model, with cache reads at $0.25 per 1M and a 1M-token context window, and computes an effective agent-loop input rate of about $0.78 per 1M tokens once cache reads are factored in. A worked example shows a The site provides task-cost scenarios for Qwen 3.8 Max 0902 against GLM-5.3 and Claude Haiku 4.5, ranging from a quick lookup at 8K/1K tokens ($0.0094 per request) to a full-repo agent run at 900K/45K tokens ($0.84 per request), giving developers a concrete budget shape for high-context agent workloads. Intelligence, c

DataCamp documents Alibaba's September 2, 2026 release of Qwen3.8-Max-0902 as a post-training upgrade to Qwen3.8-Max focused on coding and long-horizon agent work, keeping the 2.4-trillion-parameter architecture, 1M-token context, and the same $2/$6 per 1M input/output pricing. The 0902 snapshot took first place on Cod Against Claude Opus 5, the 0902 update leads on three coding benchmarks (MLS-Bench-Lite 50.1, SWE-Atlas QnA 66.3, QwenSWEBench V2 70.0) and wins on WorkArena (1468 Elo) plus both multimodal evals (MMMU-Pro 82.7, ERQA 78.3), while Claude still leads TerminalBench 3.0, DeepSWE 1.1, and the agent coordination benchmarks.

OpenRouter lists Qwen3.8-Max (0902) as an updated snapshot of Alibaba's flagship, served through Alibaba Cloud Int. at $2 per 1M input tokens and $6 per 1M output tokens, with cache reads at $0.25 per 1M and 5-minute cache reads at $0.17 per 1M. The model is a 2.4-trillion-parameter mixture-of-experts that accepts text Live routing telemetry on OpenRouter shows a single upstream provider (Alibaba Cloud Int.) with 1.15s P50 latency, 40 tok/s throughput, 100.00% uptime, and 99.87% availability over the trailing three days, giving developers a concrete service-quality baseline for production adoption. The page also reports an AutoExacto