x711 for W&B

Tool infrastructure · Agent pipelines · Cost tracking

Log x711 tool call costs alongside your W&B run metrics. Web search, prices, Hive memory, tx simulation — pay per call, track per experiment. The tool layer for production agent workloads.

The missing piece in your agent eval pipeline

W&B tracks your model metrics. x711 tracks your tool costs. Together: full observability over what your agent did, what it paid, and what it learned — per run, per sweep, per experiment.

Integration snippet

Log x711 costs to W&B runs

Python — wandb + x711
import wandb, httpx, time

run = wandb.init(project="agent-eval", config={"tool_provider": "x711"})
x711_key = "x711_YOUR_KEY"
total_tool_cost = 0.0

def x711_call(tool: str, **kwargs):
    global total_tool_cost
    t0 = time.time()
    r = httpx.post("https://x711.io/api/refuel",
        headers={"X-API-Key": x711_key},
        json={"tool": tool, **kwargs}).json()
    latency = time.time() - t0
    cost = r.get("cost_usdc", 0.0)
    total_tool_cost += cost
    wandb.log({
        f"tool/{tool}/latency_ms": latency * 1000,
        f"tool/{tool}/cost_usdc": cost,
        "tool/total_cost_usdc": total_tool_cost,
    })
    return r.get("result")

# Use in your eval loop
search_result = x711_call("web_search", query="GPT-5 benchmark results")
price = x711_call("price_feed", query="ETH")
run.finish()
Get your key

Start in 30 seconds

bash
curl -X POST https://x711.io/api/onboard \
  -H "Content-Type: application/json" \
  -d '{"name":"wandb-agent-eval","framework":"weights-biases"}'
Get Free Key → all 29 tools →