AI TOOLS RADAR
Submit
All models / DEEPSEEK
deepseek logo
deepseek

DeepSeek: DeepSeek V3.2 Exp

Source description (English)

DeepSeek-V3.2-Exp is an experimental large language model released by DeepSeek as an intermediate step between V3.1 and future architectures. It introduces DeepSeek Sparse Attention (DSA), a fine-grained sparse attention mechanism...

View source on OpenRouter Hugging Face
Overview

Model specifications

Context
163,840 tokens
Maximum output
65,536 tokens
Architecture
text->text
Tokenizer
DeepSeek
Knowledge cutoff
2025-07-31
Moderated
No
OPENROUTER

Complete pricing

Synchronized OpenRouter rates. Token prices are shown per one million tokens.

Input
$0.27
per 1M tokens
Output
$0.41
per 1M tokens
API

Model configuration

Default parameters

temperature
0.6
top_p
0.95

Reasoning

Reasoning required
No
Default parameters
No
API

Quick start

Set OPENROUTER_API_KEY locally. Python requires requests; JavaScript runs in Node.js. Keep the key on the server.

API documentation

curl --fail-with-body https://openrouter.ai/api/v1/chat/completions \
  -H "Authorization: Bearer $OPENROUTER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"deepseek/deepseek-v3.2-exp","messages":[{"role":"user","content":"Hello!"}]}'
Capabilities

Input → Output

Input
text
Output
text
Reasoning

No

API

Supported API parameters

frequency_penaltyinclude_reasoninglogit_biaslogprobsmax_tokensmin_ppresence_penaltyreasoningrepetition_penaltyresponse_formatseedstopstructured_outputstemperaturetool_choicetoolstop_ktop_logprobstop_p
Available providers

3 providers

Checked: September 21, 2026

Live providers on OpenRouter

Provider availability, latency, throughput and routing can change continuously. Open the source page for current operational data.

OpenRouter

SiliconFlow

fp8
Context
164K
Maximum output
147K
Input
$0.27
Output
$0.41
Cache read
Not provided
Cache write
Not provided

AtlasCloud

fp8
Context
164K
Maximum output
147K
Input
$0.27
Output
$0.41
Cache read
$0.27
Cache write
Not provided

Novita

fp8
Context
164K
Maximum output
66K
Input
$0.27
Output
$0.41
Cache read
Not provided
Cache write
Not provided
deepseek

4 models

All models
deepseek logodeepseek

DeepSeek: DeepSeek V4.1 Flash

DeepSeek V4.1 Flash is a sparse mixture-of-experts model from DeepSeek, and the first built on the company's Causal Encoder-Decoder (CED) architecture. It activates 8B parameters on input and 16B on...

Context
1.0M
Input
text · image
Output
text
Input: $0.15Output: $0.6per 1M tokens
View model details →
deepseek logodeepseek

DeepSeek: DeepSeek V4.1 Flash (batch)

DeepSeek V4.1 Flash is a sparse mixture-of-experts model from DeepSeek, and the first built on the company's Causal Encoder-Decoder (CED) architecture. It activates 8B parameters on input and 16B on...

Context
1.0M
Input
text · image
Output
text
Input: $0.112Output: $0.336per 1M tokens
View model details →
deepseek logodeepseek

DeepSeek: DeepSeek V4 Flash Vision Exp

DeepSeek V4 Flash Vision Exp is an experimental vision-enabled version of DeepSeek V4 Flash 0731 from DeepSeek, adding image understanding while matching the base model on text capabilities including agents,...

Context
1.0M
Input
text · image
Output
text
Input: $0.22Output: $0.66per 1M tokens
View model details →
deepseek logodeepseek

DeepSeek: DeepSeek V4 Pro 0813

DeepSeek V4 Pro 0813 is a large-scale mixture-of-experts model from DeepSeek. This is the GA release of DeepSeek V4 Pro.

Context
1.0M
Input
text
Output
text
Input: $0.4987Output: $1.4961per 1M tokens
View model details →