Skip to content
Agientry
← Back to Videos

JEV-27B-VL: Calibrated Visual Decisions for Long-Context Workflows

Loading video
Loading video

JEV-27B-VL ranks first among 20 models on the Decision Index Vision board with 77.8% accuracy over 13,695 examples and an ECE of 0.019. System 1 answers image-and-text prompts up to 256K tokens with calibrated per-option probabilities in one forward pass; System 2 handles step-by-step multimodal reasoning. Open weights under Apache 2.0.

Vision-language modelsCalibrated ConfidenceVisual Decision MakingMultimodal LLMAutoTrustModelScope
Category: llm
Author: @ModelScope2022
Date: 2026-10-10T00:00:00
Duration: 25.45s
Source: https://x.com/ModelScope2022/status/2108852576447758838
Reference: https://modelscope.ai/models/AutoTrust-Labs/JEV-27B-VL

Related Videos

Google AI Edge Foresight: Free Offline AI Note-Taking on Apple Silicon
0:47

Google AI Edge Foresight: Free Offline AI Note-Taking on Apple Silicon

@AYi_AInotes · 92 views · 2026-10-11
Google AI EdgeOn-Device AINote Taking
DeepGEMM-Ascend Brings DeepSeek Kernels to Huawei NPUs
0:15

DeepGEMM-Ascend Brings DeepSeek Kernels to Huawei NPUs

@EngMoElgaraihy · 301 views · 2026-10-05
DeepGEMMDeepGEMM-AscendDeepSeek
"Welcome to the era of Superintelligence": Fable 5.5 apparently goes live — devs report auto-routing on claude.ai, and a Superman continuous-animation demo in multiple art styles
0:40

"Welcome to the era of Superintelligence": Fable 5.5 apparently goes live — devs report auto-routing on claude.ai, and a Superman continuous-animation demo in multiple art styles

@chetaslua · 565 views · 2026-10-01
LLMAnthropicClaude
AutoTrust open-sources JEV-27B-VL: the first open-weight near-SOTA multimodal decision model — it watches Mario and decides whether to jump
0:25

AutoTrust open-sources JEV-27B-VL: the first open-weight near-SOTA multimodal decision model — it watches Mario and decides whether to jump

@AutoTrustAI · 767 views · 2026-10-01
LLMJEV-27B-VLAutoTrust AI
SpatialLM: Labeled 3D Room Layouts from Monocular Phone Video
0:51

SpatialLM: Labeled 3D Room Layouts from Monocular Phone Video

@0x0SojalSec · 292 views · 2026-10-09
Spatial Intelligence3D Scene UnderstandingVision-language models
TacZero: Training-Free Peg Insertion with Tactile VLM
0:48

TacZero: Training-Free Peg Insertion with Tactile VLM

@sports_robots · 221 views · 2026-10-07
TacZeroVision-language modelsTactile Sensing

As an Amazon Associate, we earn from qualifying purchases.