GLM-5.3 is live. Run Z.ai's latest model on Friendli Model APIs. Try it today ➜
tyy1850819840
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Model Details
Model Provider
tyy1850819840
Model Tree
Qwen/Qwen3-VL-4B-Instruct
Input Modalities
Output Modalities
Supported Functionality
| Benchmark | Metric | Score |
|---|---|---|
| RGBT-GroundBench | Acc@0.5 | 72.6788 |
| Degraded-1K | Acc@0.5 | 54.20 |
| RGB-Th-Bench | macro QAcc | 75.2481 |
| DV-1K-Balanced | Accuracy | 79.30 |
These are complete archived evaluations of the recovered checkpoint, not exact paper-table reproduction. Paper GroundBench/Degraded scores were 72.3627/48.70. Code, prompts, dataset downloads and custom-set reconstruction limitations: RelMLLM GitHub.
The model is intended for research, not safety-critical sensor-health or surveillance decisions. Reliability fields are task-conditioned evidence utilities rather than physical sensor diagnoses. See the companion repository for protocols, limitations, code, license, and attribution.