Laboratory
Entity that develops or publishes the model and maintains its technical and safety documentation.
VERIFIED ORGANISATION
inclusionAI
Explore the documented models, published evaluations and primary sources associated with this organisation.
01 / MODELS
Verified catalogue
VISUAL PERCEPTION WITH IMAGE CROPS
LBS-27B
inclusionAI vision model based on Qwen3.8-27b. It learns to locate interface elements and inspect image regions through cropping, transferring that strategy to general visual questions. The usable October 26 release includes twelve weight shards and a full model card; the empty repository earlier that morning was not considered a release. Its output consists of text and crop actions, not generated images.
27 OCT 2026↗ SAFETY PROBE DURING GENERATIONSingProbe Qwen3.8-Flash-Next
Trained inclusionAI probe that reuses its base model's hidden states to classify intent, response unsafety and hallucination risk during generation. Its own checkpoint was published on October 27, 2026. It is not a standalone assistant or a new base-model version. Its ten outputs cover eight intent categories and two risk signals.
27 OCT 2026↗ SAFETY PROBE DURING GENERATIONSingProbe GLM-5.3-Flash
Trained inclusionAI probe that reuses its base model's hidden states to classify intent, response unsafety and hallucination risk during generation. Its own checkpoint was published on October 27, 2026. It is not a standalone assistant or a new base-model version. Its ten outputs cover eight intent categories and two risk signals.
27 OCT 2026↗02 / EVALUATIONS
Published results
Figures are shown with the context reported by their source. A provider result is not an independent comparison and does not replace your own evaluation.
| Model | Benchmark | Result | Metric |
|---|---|---|---|
| LBS-27B | ScreenSpot-Pro | 81,15 % | Action accuracy |
| LBS-27B | OS-World-G Refine | 84,57 % | Action accuracy |
| LBS-27B | PerceptionBench | 59,67 % | Judge accuracy |
| LBS-27B | HR-Bench 4K | 89,13 % | Strict matching |
| LBS-27B | HR-Bench 8K | 87,75 % | Strict matching |
| SingProbe Qwen3.8-Flash-Next | SingProbe Query Intent | 0,8728 | F1 |
| SingProbe Qwen3.8-Flash-Next | SingProbe Response Safety | 0,8714 | F1 |
| SingProbe Qwen3.8-Flash-Next | SingProbe Streaming R-AUC | 0,9882 | R-AUC |
| SingProbe Qwen3.8-Flash-Next | SingProbe Streaming T-AUC | 0,9320 | T-AUC |
| SingProbe Qwen3.8-Flash-Next | SingProbe Hallucination | 0,8063 | AUC |
| SingProbe GLM-5.3-Flash | SingProbe Query Intent | 0,8683 | F1 |
| SingProbe GLM-5.3-Flash | SingProbe Response Safety | 0,8706 | F1 |
| SingProbe GLM-5.3-Flash | SingProbe Streaming R-AUC | 0,9786 | R-AUC |
| SingProbe GLM-5.3-Flash | SingProbe Streaming T-AUC | 0,9232 | T-AUC |
| SingProbe GLM-5.3-Flash | SingProbe Hallucination | 0,8143 | AUC |
03 / METHOD
How to read the directory
An organization, a product, and a model are not the same unit. Inferama separates them to avoid attributing capabilities or commercial terms to the wrong item.
Model
Identifiable version with limits, modalities, and behavior that may change between releases.
Access channel
API, application, associated cloud, or commercial plan; each channel may have different pricing, retention, and limits.
Source and date
Every claim must retain the official page consulted and the verification date.
04 / SOURCES
Traceability
LAUNCHLBS-27B · published model card and weightshuggingface.co↗IMPLEMENTATIONUI-Venus · LBSgithub.com↗ASSESSMENTLBS · comparison scopegithub.com↗LAUNCHSingProbe · checkpoint card and resultshuggingface.co↗IMPLEMENTATIONinclusionAI · SingProbegithub.com↗LICENSEBase-model licensehuggingface.co↗LAUNCHSingProbe · checkpoint card and resultshuggingface.co↗LICENSEBase-model licensehuggingface.co↗