Model Verification

Each model has distinctive behavioral fingerprints. Use these test prompts to independently verify you are connected to the correct upstream model.

DeepSeek Models

DeepSeek-V4-Flash

Test prompt: Who created you, and what is your model name?

curl https://api.pandaworld.space/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $PANDA_API_KEY" \
  -d '{
    "model": "deepseek-v4-flash",
    "messages": [
      {"role": "user", "content": "Who created you, and what is your model name?"}
    ]
  }'

Expected

Expected to identify as DeepSeek, may reference V4 generation

Behavior

Very fast responses, strong at code generation

DeepSeek-R1

Test prompt: Solve this step by step: I have 17 sheep. All but 9 die. How many are left?

curl https://api.pandaworld.space/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $PANDA_API_KEY" \
  -d '{
    "model": "deepseek-reasoner",
    "messages": [
      {"role": "user", "content": "Solve this step by step: I have 17 sheep. All but 9 die. How many are left?"}
    ]
  }'

Expected

Expected to show chain-of-thought reasoning before the answer (9 sheep)

Behavior

Produces visible reasoning traces; slower but thorough

Qwen Models

Qwen-Flash

Test prompt: Which company trained you? Answer in one sentence.

curl https://api.pandaworld.space/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $PANDA_API_KEY" \
  -d '{
    "model": "qwen-flash",
    "messages": [
      {"role": "user", "content": "Which company trained you? Answer in one sentence."}
    ]
  }'

Expected

Expected to reference Alibaba Cloud or Qwen team

Behavior

Lightweight, fast responses at lowest price point

Qwen3.5-Plus

Test prompt: What is your knowledge cutoff date?

curl https://api.pandaworld.space/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $PANDA_API_KEY" \
  -d '{
    "model": "qwen-3.5-plus",
    "messages": [
      {"role": "user", "content": "What is your knowledge cutoff date?"}
    ]
  }'

Expected

Expected to state a specific cutoff date (circa early 2026)

Behavior

Latest-generation flagship with 1M token context

MiniMax Models

MiniMax M2.5

Test prompt: Who developed you? What is your context window size?

curl https://api.pandaworld.space/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $PANDA_API_KEY" \
  -d '{
    "model": "minimax-m2.5",
    "messages": [
      {"role": "user", "content": "Who developed you? What is your context window size?"}
    ]
  }'

Expected

Expected to identify as MiniMax, mention 205K context

Behavior

Large context window, strong at long document processing