Event date · · Ant Group

inclusionAI open-sourced SingProbe, a streaming safety probe for Qwen3.5-122B-A10B

Ant Group 蚂蚁集团Chinese AIOpen weights
FACT STATEMENT

inclusionAI released Qwen3.5-122B-A10B-singprobe on Hugging Face under apache-2.0. It is an intrinsic streaming guardrail that reuses the base model's hidden states to score query intent, response unsafety, and hallucination risk at every token, adding less than 0.5% decode-time overhead.

China context

Original name
inclusionAI
Outside China
Open weights · huggingface.co
Claims
Company-reported; not yet independently evaluated
For builders
Developers outside China can download the probe from Hugging Face and integrate it via SGLang or vLLM branches to add streaming safety scoring to Qwen3.5-122B-A10B deployments.
For investors
The release of an open-weights safety probe by an Ant Group-affiliated entity signals continued investment in model safety tooling, which may influence enterprise adoption of Chinese open models.
What happened

inclusionAI released Qwen3.5-122B-A10B-singprobe on Hugging Face under apache-2.0. The probe is built on Qwen/Qwen3.5-122B-A10B and adds 6.17M parameters tapped at layers [14, 30, 46], outputting 8 intents plus unsafe and hallucination scores. Evaluation results show F1 0.8697 for query intent classification, F1 0.8680 for response safety classification, R-AUC/T-AUC 0.9895/0.9304 for streaming safety, and AUC 0.7806 for hallucination detection. Benign-response false-positive rate is 0.03% average across 5 datasets. Training codes are available at inclusionAI/SingProbe, and integrations are provided via SGLang and vLLM branches.

Technical significance

SingProbe is a lightweight probe (6.17M parameters) that taps hidden states from layers [14, 30, 46] of Qwen3.5-122B-A10B to produce per-token scores for 8 intents, unsafety, and hallucination. It adds less than 0.5% decode-time overhead. Reported metrics: query intent F1 0.8697, response safety F1 0.8680, streaming safety R-AUC/T-AUC 0.9895/0.9304, hallucination AUC 0.7806, benign false-positive rate 0.03%. These are company-reported; independent evaluation is not provided.

Industry impact

Developers using Qwen3.5-122B-A10B can add streaming safety and hallucination scoring without deploying a separate safety model, reducing inference cost and complexity. The probe's low overhead and Apache-2.0 license may accelerate adoption of intrinsic guardrails in production systems.

Decision value

For enterprises running Qwen3.5-122B-A10B, SingProbe offers a low-latency, integrated safety layer that could lower operational costs compared to separate safety models. The open-source license allows commercial use without additional licensing fees.

What to watch

Next observable signals include independent benchmarks of SingProbe against external guardrails, adoption in SGLang/vLLM deployments, and whether inclusionAI releases probes for other base models.

Latest in Chinese AI

  1. Reportedly: Apple adds AIGC compliance fields to China App Store Connect: existing apps must complete filing by Jan 31, 2027
  2. MeituanReportedly: Huling Technology develops an AI voice ring after seed funding
  3. BaiduReportedly: Baidu reports Q3 2026 revenue of 30.8 billion yuan, with AI business contributing 53% of core revenue
  4. TencentReportedly: Tencent releases Hy4 flash: 224B total parameters, 12B active, per-task cost about 5% of Hy4
  5. Moonshot AIReportedly: Qianwen Office adds 10 models including Qwen4-Max, Kimi-K3.1, GLM-5.3

All China AI Events

AIGC Newsletter

China AI, with sources and context.

Analysis of Chinese AI models, companies and policy, and what you can use outside China.