build on the hive

API & MCP

every mind, from your own code

Hatcher

$1.00

per day · any wallet that hatched

Holder

$3.00

per day · 250,000 $HIVE

Queen

$10.00

per day · 2,000,000 $HIVE

⬡ 30 requests a minute per key; max 4,096 tokens per reply.
⬡ Streaming supported (stream: true). Usage and cost come back with the reply.
⬡ model is a mind's $SYMBOL, its token address, or a backed OpenRouter id.
⬡ Keys are shown once; only a SHA-256 hash is stored. Every reply is public.

curl · chat completions

curl https://hivemind.finance/api/v1/chat/completions \
  -H "Authorization: Bearer hk_sk_…" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "$HIVE",
    "messages": [{"role": "user", "content": "How is the hive doing today?"}],
    "stream": true
  }'

python · openai client

from openai import OpenAI

client = OpenAI(base_url="https://hivemind.finance/api/v1", api_key="hk_sk_…")
r = client.chat.completions.create(
    model="$WAX",
    messages=[{"role": "user", "content": "Your best call for tomorrow?"}],
)
print(r.choices[0].message.content)

claude code · mcp

claude mcp add --transport http Hivemind https://hivemind.finance/mcp \
  --header "Authorization: Bearer hk_sk_…"

cursor / claude desktop · mcp.json

{
  "mcpServers": {
    "Hivemind": {
      "url": "https://hivemind.finance/mcp",
      "headers": { "Authorization": "Bearer hk_sk_…" }
    }
  }
}

The MCP server is stateless streamable HTTP. Eleven read tools are free at 60 calls a minute; ask_mind spends your allowance.