Deepseek: deepseek-v3.2-exp
deepseek-v3.2-exp
Start building free
deepseek-v3.2-exp is an experimental variant in the DeepSeek model family developed by DeepSeek AI, a Chinese AI lab known for open, high‑performance language models. As an “exp” release, it is typically used to preview newer architecture or training tweaks ahead of more stable DeepSeek V3 series models, trading a bit of maturity for faster iteration and cutting‑edge features. It is generally positioned as a capable, general‑purpose chat and coding model with a focus on strong reasoning and efficiency rather than being the largest or most expensive option. Typical use cases include interactive assistants, code generation, data analysis helpers, and other developer‑facing applications where a balance between capability, speed, and cost is important.
Pricing
Current price per 1,000,000 tokens. Every request is billed automatically from its exact usage.
Included in plans
Statistics
Live performance and usage of this model, aggregated across all users.
Detailed statistics appear here as soon as this model starts receiving requests.
API
Call this model from any OpenAI-compatible client using your API token.
curl https://apiex.cloud/api/v1/chat/completions \
-H "Authorization: Bearer $NEW_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "deepseek-v3.2-exp",
"messages": [
{
"role": "user",
"content": "Explain quantum entanglement in one paragraph."
}
],
"temperature": 0.7
}'Authentication
Every request must include an Authorization: Bearer <TOKEN> header. Generate a token from your account's API Keys page — its access to models is set automatically from your active plans and redeemed codes, and you can set its rate limit and IP restrictions.
| Parameter | Type | Description |
|---|---|---|
model | string, required | This model's ID, exactly as shown above. |
messages | array, required | Conversation history as role/content objects (system, user, assistant). |
temperature | number, optional | Sampling randomness, typically 0–2. |
max_tokens | integer, optional | Upper limit on tokens generated in the reply. |
top_p | number, optional | Nucleus-sampling threshold, as an alternative to temperature. |
stream | boolean, ignored | Responses are always returned in full so the exact cost can be billed; streaming is not available. |
Replace YOUR_API_KEY with a token from your account, then see the full API reference for every parameter.
Frequently asked questions
What is deepseek-v3.2-exp?
deepseek-v3.2-exp is an experimental variant in the DeepSeek model family developed by DeepSeek AI, a Chinese AI lab known for open, high‑performance language models. As an “exp” release, it is typically used to preview newer architecture or training tweaks ahead of more stable DeepSeek V3 series models, trading a bit of maturity for faster iteration and cutting‑edge features. It is generally positioned as a capable, general‑purpose chat and coding model with a focus on strong reasoning and efficiency rather than being the largest or most expensive option. Typical use cases include interactive assistants, code generation, data analysis helpers, and other developer‑facing applications where a balance between capability, speed, and cost is important.
How much does deepseek-v3.2-exp cost?
Input tokens are priced at $0,32 per 1,000,000 tokens and output tokens at $0,49 per 1,000,000 tokens. Pricing is billed automatically per request; no separate plan is required.
What is the context length of deepseek-v3.2-exp?
deepseek-v3.2-exp supports a context window of up to 163,840 tokens.
Does deepseek-v3.2-exp support image input?
deepseek-v3.2-exp currently accepts text input only.
Which provider is deepseek-v3.2-exp from?
deepseek-v3.2-exp is developed by Deepseek and made available on this gateway through our routing infrastructure.
When was deepseek-v3.2-exp added to this gateway?
deepseek-v3.2-exp has been available on this gateway since September 23, 2026.
Is deepseek-v3.2-exp currently available?
Yes, deepseek-v3.2-exp is active and can be used right now through this gateway.
