Alibaba Prices Qwen3.8-Flash API At $0.16 Per Million Tokens, Cutting Inference Costs For 125B-Parameter Model

by
Alisa Davidson

Revealed: August 26, 2026 at 10:41 am Up to date: August 26, 2026 at 10:41 am

by Anastasiia O

Edited and fact-checked:
August 26, 2026 at 10:41 am

To enhance your local-language expertise, typically we make use of an auto-translation plugin. Please be aware auto-translation is probably not correct, so learn original article for exact info.

The Qwen Staff has launched Qwen3.8-Flash-Subsequent, an open-weight multimodal mixture-of-experts mannequin that serves as an early architectural preview for the upcoming Qwen4 collection. The mannequin balances substantial capability with distinctive price effectivity, that includes 125 billion complete parameters and a further 51 billion N-gram embedding parameters whereas activating solely 6 billion parameters per token.

The discharge follows the precedent established by Qwen3-Subsequent, which launched hybrid structure designs later adopted throughout the Qwen3.5 via Qwen3.8 households. Qwen3.8-Flash-Subsequent can be accessible via the QwenCloud API at a price of $0.16 per million enter tokens and $0.47 per million output tokens, positioning it as a extremely aggressive possibility for high-volume functions, coding assistants, and enterprise agentic workflows.

Benchmark outcomes point out robust efficiency throughout software program engineering and autonomous agent duties. The mannequin achieved 58.7% on DeepSWE 1.1, 62.5% on SWE-bench Professional, and 81.0% on the multilingual variant. In long-horizon workplace automation measured by CoWorkBench, it scored 73.9%, surpassing each Qwen3.7-Plus and Claude-Opus-4.6. 

Common reasoning capabilities stay sturdy, with scores of 91.7% on GPQA Diamond and 91.9% on LiveCodeBench v6. Multimodal competence is equally stable, with 84.5% on AndroidWorld and 76.6% on LVBench for lengthy video understanding. Native context size reaches 262,144 tokens, extensible to 1 million tokens by way of YaRN.

Architectural Improvements and Developer Integration

The structure introduces 4 systematic upgrades. For consideration, the mannequin combines Gated DeltaNet with Qwen Sparse Consideration, compressing historic context effectively whereas retrieving related info via micro-block indexing moderately than token-level processing. This design achieves as much as 7.6× prefill speedup at a million tokens. 

The Gated Residual mechanism expands the residual stream into 4 parallel branches with dynamic gating, enhancing cross-layer info stream and coaching stability whereas supporting FP8 storage for diminished reminiscence visitors. N-gram Embedding provides capability via local-context lookups that require minimal computation and might be asynchronously prefetched from host reminiscence. Coaching employs the Muon optimizer with refined orthogonalization and parameter-splitting methods, enabling steady convergence at bigger batch sizes with out conventional warmup procedures.

The mannequin weights can be found on HuggingFace and ModelScope, with API entry via QwenCloud supporting OpenAI-compatible Chat Completions and Anthropic-compatible protocols. Builders can combine the mannequin into current workflows via Claude Code, OpenAI Codex, Qoder CLI, Qwen Code, and OpenClaw, with reasoning effort configurable throughout low, medium, and xhigh ranges. An official manufacturing launch with built-in instruments and default one-million-token context is predicted to observe shortly.

Alisa, a devoted journalist on the MPost, focuses on crypto, AI, investments, and the expansive realm of Web3. With a eager eye for rising developments and applied sciences, she delivers complete protection to tell and interact readers within the ever-evolving panorama of digital finance.

• 
  
  
  

More articles