DeepSeek‑V3.2‑Exp Thinking API

deepseek/deepseek-thinking-v3.2-exp
DeepSeek-V3.
Context
128K tokens
Input
$0.371358 / 1M tokens
Output
$0.563914 / 1M tokens
Released
Oct 3, 2025

How to use DeepSeek‑V3.2‑Exp Thinking API

Install any OpenAI-compatible SDK, point it at api.aimlapi.com/v1, and set the model to deepseek/deepseek-thinking-v3.2-exp.
import requests

r = requests.post(
    "https://api.aimlapi.com/v1/chat/completions",
    headers={"Authorization": "Bearer " + AIMLAPI_KEY},
    json={
      "model": "deepseek/deepseek-thinking-v3.2-exp",
      "messages": [
        {
          "role": "user",
          "content": "Hello!"
        }
      ]
    },
)
print(r.json())
const r = await fetch("https://api.aimlapi.com/v1/chat/completions", {
  method: "POST",
  headers: {
    Authorization: `Bearer ${process.env.AIMLAPI_KEY}`,
    "Content-Type": "application/json",
  },
  body: JSON.stringify({
    "model": "deepseek/deepseek-thinking-v3.2-exp",
    "messages": [
      {
        "role": "user",
        "content": "Hello!"
      }
    ]
  }),
});
console.log(await r.json());
curl -X POST https://api.aimlapi.com/v1/chat/completions \
  -H "Authorization: Bearer $AIMLAPI_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"deepseek/deepseek-thinking-v3.2-exp","messages":[{"role":"user","content":"Hello!"}]}'

OpenAI-compatible — swap the base URL and it works with your existing SDK.

DeepSeek‑V3.2‑Exp Thinking API Pricing

TypePrice
Input
$0.371358 / 1M tokens
Output
$0.563914 / 1M tokens
Cached input
$0.371358 / 1M tokens

DeepSeek‑V3.2‑Exp Thinking Benchmarks

BenchmarkScoreWhat it measuresSourceRetrieved
Intelligence
16.6
Composite score across standardised reasoning, knowledge and problem-solving evaluations, measured independently by Artificial AnalysisSourceSeptember 12, 2026
Math
87.7
Composite score across standardised mathematics evaluations, measured independently by Artificial AnalysisSourceSeptember 12, 2026
MMLU-Pro
85.0
Multi-discipline knowledge + reasoning (harder MMLU)SourceSeptember 22, 2026
GPQA Diamond
79.9
Google-proof graduate science questions (hardest subset)SourceSeptember 22, 2026
Humanity's Last Exam
19.8
Expert-level questions across many domainsSourceSeptember 22, 2026
LiveCodeBench
74.1
Contamination-free competitive programming problemsSourceSeptember 22, 2026
AIME 2025
89.3
Competition mathematics (AIME), 2025SourceSeptember 22, 2026
SWE-bench Verified
67.8
Resolving verified real GitHub issuesSourceSeptember 22, 2026

DeepSeek‑V3.2‑Exp Thinking vs other models

ModelInputOutputContextBest for
$0.371358 / 1M tokens
$0.563914 / 1M tokens
128K tokens
Coding + agents
$6.5 / 1M tokens
$39 / 1M tokens
1M tokens
Reasoning + agents
$2.6 / 1M tokens
$13 / 1M tokens
1M tokens
Balanced coding + agents
$3.9 / 1M tokens
$19.5 / 1M tokens
1M tokens
Long-context, multimodal & agentic workflows
$0.65 / 1M tokens
$3.9 / 1M tokens
1M tokens
Reasoning + agents

Frequently asked questions

DeepSeek‑V3.2‑Exp Thinking has a 128,000 tokens context window and can return up to 64,000 tokens.

DeepSeek‑V3.2‑Exp Thinking takes text as input and returns text.

Use deepseek/deepseek-thinking-v3.2-exp as the model id. Requests go to https://api.aimlapi.com/v1/chat/completions.

DeepSeek‑V3.2‑Exp Thinking is priced at input $0.371358 / 1M tokens, output $0.563914 / 1M tokens, cached input $0.371358 / 1M tokens.

Yes, DeepSeek‑V3.2‑Exp Thinking can stream responses as they are generated.

DeepSeek‑V3.2‑Exp Thinking was built by DeepSeek.

Start building with DeepSeek‑V3.2‑Exp Thinking

Get API Key
1000+ models, one API.