MODEL DIFF

What changed in DeepSeek-V4-Flash-Vision-Exp?

Compare any two evidence snapshots. Differences are source-derived and shown field by field without a quality judgment.

DeepSeek-V4-Flash-Vision-Exp: latest recorded evidence change

Observed on : 5 changed evidence field(s). These are recorded source observations, not claims that the model is better, safer or faster.

Repository versions: 6821d6ad3681a4b1 → 6821d6ad3681a4b1

license.commercial_use_label

BeforeCommercial use allowed under MIT conditions
AfterCommercial use allowed under mit conditions

license.name

BeforeMIT
Aftermit

runtime.signals (file entries)

Before3
After4

training.recipe_status

Beforetraining-or-recipe-links-present
Aftertraining-described-in-model-card

weights.precision (file entries)

Before2
After0

Showing 5 of 5 changed field(s). Use the interactive comparison below for the complete evidence history.

Compare any two retained snapshots

→

Loading evidence ledger…