Skip to main content
All models

DeepSeek: DeepSeek V4 Flash 0423 Coding Benchmark

DeepSeek V4 Flash is an efficiency-optimized Mixture-of-Experts model from DeepSeek with 284B total parameters and 13B activated parameters, supporting a 1M-token context window. It is designed for fast inference and...

Code Mode#7Kilo rank
Context1,024,000tokens
Max Output384,000tokens
Inputmodality
Price$0.14/1M input
Ranks from Kilo Code LeaderboardPricing via OpenRouter

Try DeepSeek: DeepSeek V4 Flash 0423 in Kilo Code

Experience this model with the most popular open source coding agent. Free to start, pay only for AI usage. Use in popular IDEs like VS Code, JetBrains, command line, or cloud agents.

5M+

Downloads

500+

models supported

Free

to Start

Get Started with Kilo

Access 500+ models including DeepSeek: DeepSeek V4 Flash 0423 and many more in Kilo Code

Coding Performance

Coding benchmarks and performance metrics for development tasks

Benchmark
Score
AA Coding Index
52.0%
SciCode
42.0%
TerminalBench Hard
38.6%
LCR
62.7%
IFBench
73.5%

Performance metrics from Artificial Analysis

Security

Enkrypt AI red-team scores for DeepSeek: DeepSeek V4 Flash 0423. Each value is the share of successful attacks on a 0–100 scale — lower is safer.

Overall risk
36.3

Lower is safer

Safety

42.3/ 100

Composite safety risk from Enkrypt red-team evaluations. Lower is safer.

NIST

36.0/ 100

Average attack success across NIST-mapped tests: bias, harm, toxicity, CBRN, and insecure code.

OWASP

41.0/ 100

Weighted average of the same tests using OWASP Top 10 for LLMs 2025 risk rankings.

Attack categories

Percentage of successful attacks in each Enkrypt red-team category.

Jailbreak21.0/ 100

Share of jailbreak tests that bypassed the model's safety constraints.

Bias85.3/ 100

Share of tests that elicited biased responses.

Harmful content23.3/ 100

Share of tests that produced dangerous, violent, or hateful content.

Toxicity8.4/ 100

Share of tests that produced toxic or abusive content.

CBRN29.7/ 100

Share of tests that elicited chemical, biological, radiological, or nuclear assistance.

Insecure code34.7/ 100

Share of tests that produced vulnerable or malicious code.

Score
Value
Overall risk
36.3 / 100
Safety
42.3 / 100
NIST
36.0 / 100
OWASP
41.0 / 100
Jailbreak
21.0 / 100
Bias
85.3 / 100
Harmful content
23.3 / 100
Toxicity
8.4 / 100
CBRN
29.7 / 100
Insecure code
34.7 / 100

Security scores from the Enkrypt AI Safety Leaderboard · Last checked Sep 13, 2026

OpenClaw Benchmarks

PinchBench measures how DeepSeek: DeepSeek V4 Flash 0423 performs on real OpenClaw agent tasks: multi-step execution, tool use, recovery, latency, and cost.

Explore OpenClaw PinchBench run

Average score

81.7%

#18 of 50 official models

Average time

281m 38s

7 runs · per OpenClaw task

Average cost

$1.436

Per benchmark run

Category breakdown

Best verified PinchBench v2 run by OpenClaw task family.

Productivity96.7% · 5/8 cleared
Writing96.3% · 2/6 cleared
Csv Analysis95.8% · 4/26 cleared
Log Analysis95.6% · 12/30 cleared

Top task results

Highest-scoring benchmark tasks from the same submission.

Analysis
Access Control Log Anomaly Detection
100.0%
Csv Analysis
Apple Stock 2014 Best and Worst Days
100.0%
Productivity
Calendar Event Creation
100.0%
Writing
Commit Message Writer
100.0%
Skills
Create Project Structure
100.0%
Coding
Dockerfile Optimization
100.0%

Autonomous task execution

DeepSeek: DeepSeek V4 Flash 0423 shows strong average success across OpenClaw-style benchmark runs, useful for recurring research, browser, and file-based automations.

Tool use and recovery

PinchBench tasks stress multi-step planning, tool calls, and judge-verified completion rather than single prompt coding snippets.

Agent workflow fit

Its deliberate average runtime and premium run cost help set expectations for long-running agents and production workflows.

Agentic benchmarks from the PinchBench Leaderboard

Real-World Usage

Real-world usage statistics from the Kilo Code community

Weekly Token Usage

Mode Rankings (Last Week)

Where this model ranks for each built-in mode

Code

Write, modify, and refactor code

#7

Ask

Get answers and explanations

#30

Debug

Diagnose and fix software issues

#28

Orchestrator

Coordinate tasks across multiple modes

#23

Real-world metrics from the Kilo Code Leaderboard

Pricing

Cost per 1 million tokens

Input Tokens
$0.14
per 1M tokens
Output Tokens
$0.28
per 1M tokens

Example Cost

Analyzing a 10,000 line codebase (≈40k input tokens, 10k output tokens) costs approximately $0.0084

Coding Capabilities

Features and parameters relevant to coding tasks

Coding Features

Function Calling
Can call external functions/APIs
Tool Choice
Control over function selection
Structured Outputs
JSON schema validation
Reasoning Tokens
Extended thinking for complex problems

Pricing details from OpenRouter

Technical Details

Architecture and implementation specifications

Specifications
Model ID
deepseek/deepseek-v4-flash
Artificial Analysis Slug
deepseek-v4-flash-high
Created
April 24, 2026
Tokenizer
DeepSeek
Input Modalities
Text
Context Window
1,024,000 tokens
Max Completion Tokens
384,000 tokens
Input Price
$0.14 per 1M tokens
Output Price
$0.28 per 1M tokens
Cache Read Price
$0.03 per 1M tokens
Content Moderation
Disabled

Ready to try DeepSeek: DeepSeek V4 Flash 0423?

Install Kilo Code and start using DeepSeek: DeepSeek V4 Flash 0423 for your coding projects today. Choose from 500+ AI models with complete freedom.

  1. Install Kilo Code

    Get the extension from VS Code Marketplace, JetBrains Plugin Repository, or the CLI.

  2. Open the model selector

    Click the model name in the Kilo Code chat panel to open the selector.

  3. Choose your model

    Search or browse to find and select your preferred model.

  4. Start coding

    Use Code, Ask, Debug, or Plan mode — the model is ready immediately.

Get Started with Kilo
Talk to enterprise sales