DeepSeek logo

DeepSeek-V4-Flash-Vision-Exp

Multimodal
DeepSeek

DeepSeek-V4-Flash-Vision-Exp is an experimental multimodal vision-understanding model on the DeepSeek API (`model='deepseek-v4-flash-vision-exp'`). It accepts text and image inputs and produces text.

Key Specifications

Parameters
-
Context
1.0M
Release Date
August 21, 2026
Average Score
50.4%

Timeline

Key dates in the model's history
Announcement
August 21, 2026
Last Update
August 29, 2026
Today
August 30, 2026

Technical Specifications

Parameters
-
Training Tokens
-
Knowledge Cutoff
-
Family
-
Capabilities
MultimodalZeroEval

Pricing & Availability

Input (per 1M tokens)
$0.22
Output (per 1M tokens)
$0.66
Max Input Tokens
1.0M
Max Output Tokens
393.2K
Supported Features
Function CallingStructured OutputCode ExecutionWeb SearchBatch InferenceFine-tuning

Benchmark Results

Model performance metrics across various tests and benchmarks

Other Tests

Specialized benchmarks
Agents' Last Exam
DeepSeek Harness minimal mode; max reasoning effort; temperature 1.0; top_p 0.95. Changelog notes that the text model DeepSeek-V4-Flash ignores multimodal elements on ApexBench and Agents' Last Exam; vision-exp score is reported separately.Self-reported
27.3%
AutomationBench
Public set; DeepSeek Harness minimal mode; max reasoning effort; temperature 1.0; top_p 0.95.Self-reported
25.7%
DeepSWE
DeepSeek Harness minimal mode; max reasoning effort; temperature 1.0; top_p 0.95.Self-reported
59.3%
DSBench-Hard
Internal difficult coding-agent test set; DeepSeek Harness minimal mode; max reasoning effort; temperature 1.0; top_p 0.95.Self-reported
63.6%
NL2Repo
DeepSeek Harness minimal mode; max reasoning effort; temperature 1.0; top_p 0.95.Self-reported
57.7%
Terminal-Bench 2.1
DeepSeek Harness minimal mode; max reasoning effort; temperature 1.0; top_p 0.95.Self-reported
83.9%
ZEROBench
Pass@5; self-reported on DeepSeek API changelog 2026-08-21.Self-reported
35.0%

License & Metadata

License
Not specified
Announcement Date
August 21, 2026
Last Updated
August 29, 2026

Compare DeepSeek-V4-Flash-Vision-Exp

All comparisons

Similar Models

All Models

Recommendations are based on similarity of characteristics: developer organization, multimodality, parameter size, and benchmark performance. Choose a model to compare or go to the full catalog to browse all available AI models.