Cloudflare reportedly launches open-source multimodal decision model Clef based on Qwen, IT Home reports
Model: Qwen3.8-27B · availability, license and releases
Reported by IT Home · not yet confirmed by the company or a second independent outlet. We update this page when it is.
Cloudflare launched Clef and Clef-flash, open-source multimodal decision models based on Qwen3.8-27B and Qwen3.5-9B, IT Home reported on October 2, 2026. Clef leads the Jev Decision Index 0.2.1 benchmark and extends context window to 64k.
China context
- Original name
- Clef
- Outside China
- Not stated in the sources yet
- Claims
- Company-reported; not yet independently evaluated
- For builders
- Developers can test Clef via Jev-API compatibility, but must verify if model weights are publicly available and if API access is provided outside China.
- For investors
- Cloudflare's adoption of Qwen as a base model indicates growing commercial use of Chinese open-weight models by global infrastructure providers, but the extent of this partnership and revenue impact is unconfirmed.
Translated from Chinese. Quotes and facts link to the original sources.
Cloudflare introduced Clef, an open-source multimodal decision model built on Alibaba's Qwen, with two variants: Clef (Qwen3.8-27B) and Clef-flash (Qwen3.5-9B). The models are fully compatible with Jev-API and lead the Jev Decision Index 0.2.1 benchmark. Clef natively includes a vision encoder for image classification, unlike Jev's text-only support, and expands the context window to 64k tokens (vs. Jev's 32k). Cloudflare also released a reinforcement learning product for custom fine-tuning.
Clef uses Qwen as its base model: Clef is based on Qwen3.8-27B and Clef-flash on Qwen3.5-9B. It adds a native vision encoder for direct image input and classification, and increases context window to 64k tokens, enabling richer system state input for decision-making.
Developers using Jev for decision-making tasks can now access a multimodal, open-source alternative with a larger context window and vision support, potentially reducing reliance on text-only models and enabling more complex decision workflows.
Cloudflare's Clef offers enterprises a customizable decision model with vision capabilities and a 64k context window, compatible with existing Jev-API workflows, and a reinforcement learning product for fine-tuning to specific business scenarios.
Verification needed: whether Clef weights are publicly downloadable and whether the models are accessible outside China via Cloudflare's global infrastructure. Watch for Cloudflare's official model release page and any independent benchmark evaluations.