Call DeepSeek V3 and the R1 reasoner for free. The keys are OpenAI-compatible — copy one, set the base URL, and get strong reasoning and coding at zero cost. No DeepSeek account, no sign-up, no card.
A working API key that reaches DeepSeek V3 (deepseek-chat)
and DeepSeek R1 (deepseek-reasoner) through an
OpenAI-compatible gateway. DeepSeek is known for excellent price-to-performance —
and here it's free. V3 is a fast general model; R1 adds explicit step-by-step
reasoning for math, logic, and hard coding tasks.
https://aiapiv2.pekpik.com/v1.deepseek-chat (V3) or deepseek-reasoner (R1).curl https://aiapiv2.pekpik.com/v1/chat/completions \
-H "Authorization: Bearer YOUR_FREE_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "deepseek-chat",
"messages": [{"role": "user", "content": "Refactor this function for readability"}]
}'from openai import OpenAI
client = OpenAI(
api_key="YOUR_FREE_KEY",
base_url="https://aiapiv2.pekpik.com/v1",
)
resp = client.chat.completions.create(
model="deepseek-reasoner",
messages=[{"role": "user", "content": "If 3x + 7 = 22, solve for x and explain."}],
)
print(resp.choices[0].message.content)Yes. Keys are shared publicly with a pre-funded budget — no credit card, no sign-up. Because they're shared, a key may run out of budget; grab another from the table or come back later.
DeepSeek V3 via deepseek-chat and R1 via deepseek-reasoner. Both are OpenAI-compatible — call them through the OpenAI SDK by changing the base URL.
deepseek-chat (V3) is fast and general-purpose. deepseek-reasoner (R1) spends extra compute on step-by-step reasoning — stronger for math, logic, and hard coding.
That's normal for shared keys: it likely ran out of budget. Grab another from the table, or use Smart mode in the playground, which auto-routes to an available model.