Skip to main content
DeepSeekApr 24, 2026

DeepSeek V4 Flash 0423

deepseek-v4-flashPrivateOpen weights

DeepSeek V4 Flash is an efficiency-optimized 284B-parameter Mixture-of-Experts model with 13B active parameters and a 1M-token context window. Tuned for fast inference and high-throughput workloads while maintaining strong reasoning and coding performance.

$0.138/$0.275
per 1M in / out
1M
context window
256K
max output
Also available end-to-end encrypted$0.182 / $0.373 per 1M
How it comparesamong 143 text models
GPQA Diamondhigher than 59%87.4%
Blended pricecheaper than 95%$0.17
Context windowlarger than 42%1M
Max outputlarger than 96%256K

Further right is better. Hover a dot for its model.

Time to first token— 
Output speed— 
Uptime— 

Last 24 hours of Venice traffic.How it’s measured

Capabilities

What the model takes in, returns and supports through the API.

InText
OutText

Not supportedimage input, video input, audio input, X search and log probabilities

Benchmarks

GPQA Diamond
87.4%56th of 139 · none scores higher for less

Specs

Model ID
deepseek-v4-flash
Endpoint
POST /api/v1/chat/completions
Context window
1,000,000 tokens
Max output
262,144 tokens
Reasoning
none · low · medium · high · default low
Precision
Not disclosed
Privacy
Private
Added
Apr 24, 2026

Compare

ModelInputOutputGPQA DiamondContextMax outputFirst tokenSpeed
DeepSeek V4 Flash 0423$0.138$0.27587.4%1M256K——
Google Gemma 4 31B Instruct$0.12$0.3686.9%256K16K——
Google Gemma 4 26B A4B Instruct$0.13$0.4071.2%256K16K——
GLM 4.7 Flash Heretic$0.07$0.4035.4%200K24K——
GPT-6 Luna$0.10$0.5088.9%1.05M128K——

Other versions

  1. DeepSeek V4 Flash 0423Apr 2026
  2. DeepSeek V4 Flash 0731Jul 2026
  3. DeepSeek V4.1 FlashSep 2026

Pricing

Billed per token in USD. Reasoning tokens bill as output.

Input$0.138per 1M tokens
Cached input$0.02880% off input
Output$0.275per 1M tokens

Estimate a workload

$0.0004per request
$4.13for 10,000 requests

Blended price against similar models

DeepSeek V4 Flash 0423$0.1723

Per 1M tokens at three input tokens for each output token.

Quickstart

OpenAI-compatible: point any OpenAI SDK at the Venice base URL with a Venice API key.

curl https://api.venice.ai/api/v1/chat/completions \
-H "Authorization: Bearer $VENICE_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "deepseek-v4-flash",
"messages": [{ "role": "user", "content": "Explain TEE attestation in two sentences." }],
"reasoning_effort": "low"
}'
Base URL https://api.venice.ai/api/v1

Parameters

ParameterValuesNotes
modeldeepseek-v4-flashRequired
messagestextContent parts the model reads
max_completion_tokensUp to 262,144Counts reasoning tokens
reasoning_effortnone | low | medium | highDefault low
toolsFunction definitionsWith tool_choice
response_format{ type: "json_schema" }Output that matches a schema
streamtrue | falseServer-sent events
venice_parameters.enable_web_searchauto | on | offWeb results and citations