Google GenAI

Google Gen AI provides the Gemini family of large language models through the Google Gen AI Python SDK and the Google Gen AI Go SDK. In Python, Arize AX captures every Gemini API call — chat completions, tool calls, and token usage — via the openinference-instrumentation-google-genai package, which also covers calls routed through Vertex AI when the SDK is configured against Vertex. In Go, there is no OpenInference instrumentor for the GenAI SDK yet, so you wrap each call in an LLM span and set attributes from the OpenInference Go semantic conventions package.

Prerequisites

Python 3.10+ or Go 1.25+
An Arize AX account (sign up)
A GEMINI_API_KEY from Google AI Studio

Launch Arize AX

Sign in to your Arize AX account.
From Space Settings, copy your Space ID and API Key. You will set them as ARIZE_SPACE_ID and ARIZE_API_KEY below.

Install

pip install arize-otel openinference-instrumentation-google-genai google-genai

go get \
  github.com/Arize-ai/arize-otel-go \
  github.com/Arize-ai/openinference/go/openinference-semantic-conventions \
  go.opentelemetry.io/otel \
  google.golang.org/genai

Configure credentials

export ARIZE_SPACE_ID="<your-space-id>"
export ARIZE_API_KEY="<your-api-key>"
export ARIZE_PROJECT_NAME="google-genai-tracing-example"
export GEMINI_API_KEY="<your-gemini-api-key>"

Setup tracing

# instrumentation.py
import os

from arize.otel import register
from openinference.instrumentation.google_genai import GoogleGenAIInstrumentor

tracer_provider = register(
    space_id=os.environ["ARIZE_SPACE_ID"],
    api_key=os.environ["ARIZE_API_KEY"],
    project_name=os.environ["ARIZE_PROJECT_NAME"],
)

GoogleGenAIInstrumentor().instrument(tracer_provider=tracer_provider)
print("Arize AX tracing initialized for Google GenAI.")

// main.go
//
// Go integrations live in a single file — the tracer setup, the
// manually instrumented Gemini call, and the output all sit inside main().
// The `Run Google GenAI` step below is just `go run main.go`.
package main

import (
    "context"
    "fmt"
    "log"
    "os"
    "time"

    arizeotel "github.com/Arize-ai/arize-otel-go"
    oi "github.com/Arize-ai/openinference/go/openinference-semantic-conventions"
    "go.opentelemetry.io/otel"
    "go.opentelemetry.io/otel/attribute"
    "go.opentelemetry.io/otel/codes"
    "google.golang.org/genai"
)

const modelName = "gemini-3.8-flash"

func main() {
    ctx := context.Background()

    projectName := os.Getenv("ARIZE_PROJECT_NAME")
    if projectName == "" {
        projectName = "google-genai-tracing-example"
    }

    // Register reads ARIZE_SPACE_ID / ARIZE_API_KEY from the environment.
    tp, err := arizeotel.Register(ctx, arizeotel.Options{
        ProjectName: projectName,
    })
    if err != nil {
        log.Printf("register tracer: %v", err)
        return
    }
    defer func() {
        shutdownCtx, cancel := context.WithTimeout(
            context.Background(), 5*time.Second,
        )
        defer cancel()
        _ = tp.Shutdown(shutdownCtx)
    }()

    fmt.Println("Arize AX tracing initialized for Google GenAI.")

    // The client reads GEMINI_API_KEY from the environment.
    client, err := genai.NewClient(ctx, &genai.ClientConfig{
        Backend: genai.BackendGeminiAPI,
    })
    if err != nil {
        log.Printf("genai client: %v", err)
        return
    }

    prompt := "Why is the ocean salty? Answer in two sentences."

    // There is no OpenInference instrumentor for the Go GenAI SDK, so wrap
    // the call in an LLM span and set OpenInference attributes by hand.
    ctx, span := otel.Tracer(projectName).Start(ctx, "GenerateContent")
    defer span.End()
    span.SetAttributes(
        attribute.String(oi.OpenInferenceSpanKind, oi.SpanKindLLM),
        attribute.String(oi.LLMProvider, oi.LLMProviderGoogle),
        attribute.String(oi.LLMModelName, modelName),
        attribute.String(oi.InputValue, prompt),
        attribute.String(oi.InputMimeType, oi.MimeTypeText),
        attribute.String(oi.LLMInputMessageRoleKey(0), "user"),
        attribute.String(oi.LLMInputMessageContentKey(0), prompt),
    )

    resp, err := client.Models.GenerateContent(
        ctx, modelName, genai.Text(prompt), nil,
    )
    if err != nil {
        span.RecordError(err)
        span.SetStatus(codes.Error, err.Error())
        log.Printf("gemini: %v", err)
        return
    }

    text := resp.Text()
    span.SetAttributes(
        attribute.String(oi.OutputValue, text),
        attribute.String(oi.OutputMimeType, oi.MimeTypeText),
        attribute.String(oi.LLMOutputMessageRoleKey(0), "model"),
        attribute.String(oi.LLMOutputMessageContentKey(0), text),
    )
    if usage := resp.UsageMetadata; usage != nil {
        // Gemini Flash models think before answering. Count the thinking
        // tokens as completion tokens and also record them as reasoning.
        completion := usage.CandidatesTokenCount + usage.ThoughtsTokenCount
        span.SetAttributes(
            attribute.Int(oi.LLMTokenCountPrompt,
                int(usage.PromptTokenCount)),
            attribute.Int(oi.LLMTokenCountCompletion, int(completion)),
            attribute.Int(oi.LLMTokenCountCompletionDetailsReasoning,
                int(usage.ThoughtsTokenCount)),
            attribute.Int(oi.LLMTokenCountTotal,
                int(usage.TotalTokenCount)),
        )
    }
    span.SetStatus(codes.Ok, "")

    fmt.Println(text)
}

Go SDK The span is created by hand, so it captures only the attributes you set. To trace streaming calls, tool calls, or chat sessions, add the matching attributes from the OpenInference semantic conventions. See Manual instrumentation for the span patterns.

Run Google GenAI

# example.py

# Importing instrumentation first ensures tracing is set up
# before `google.genai` is imported.
from instrumentation import tracer_provider

from google import genai

# The client reads GEMINI_API_KEY from the environment.
client = genai.Client()

response = client.models.generate_content(
    model="gemini-3.8-flash",
    contents="Why is the ocean salty? Answer in two sentences.",
)

print(response.text)

# The full example lives in main.go above.
go run main.go

Expected output

Arize AX tracing initialized for Google GenAI.
The ocean is salty because rivers continuously dissolve mineral salts from rocks and soil and carry them to the sea, where they accumulate over millions of years. Water leaves the ocean through evaporation but the salts remain, steadily concentrating until reaching today's roughly 3.5% salinity.

Verify in Arize AX

Open your Arize AX space and select project google-genai-tracing-example.
You should see a new trace within ~30 seconds containing a GenerateContent LLM span with the prompt, response, and token usage attached.
If no traces appear, see Troubleshooting.

Check from the skill, CLI, or SDK

Confirm spans are actually reaching your Arize AX project. Use whichever fits your workflow — the skill and CLI work for any framework; the SDK check is shown for each language.

Arize skill (agent)
AX CLI
SDK

Install the Arize Skills plugin and let your coding agent check for you:

npx skills add Arize-ai/arize-skills

Then prompt your agent:

Use the arize-trace skill to export and analyze recent traces from my project. Confirm spans are arriving, and summarize any errors or latency issues.

Export recent spans for your project — any rows mean traces are landing:

ax spans export "$ARIZE_PROJECT_NAME" --space "$ARIZE_SPACE_ID" \
  --limit 5 --stdout | jq 'length'

A non-zero count confirms spans reached Arize AX. Run ax auth login first if you have not authenticated. See the ax spans reference.

Query the project’s spans and check that at least one came back.

import os
from arize import ArizeClient

client = ArizeClient(api_key=os.environ["ARIZE_API_KEY"])
resp = client.spans.list(
    project=os.environ["ARIZE_PROJECT_NAME"],
    space=os.environ["ARIZE_SPACE_ID"],
    limit=5,
)
count = len(resp.spans)
print(
    f"{count} span(s) found" if count else "No spans yet — recheck setup"
)

// Reads ARIZE_API_KEY from the environment.
import { listSpans } from "@arizeai/ax-client";

const { data: spans } = await listSpans({
  project: process.env.ARIZE_PROJECT_NAME!,
  space: process.env.ARIZE_SPACE_ID!,
  limit: 5,
});
const count = spans.length;
console.log(
  count ? `${count} span(s) found` : "No spans yet — recheck setup",
);

client, err := arize.NewClient(
    arize.Config{APIKey: os.Getenv("ARIZE_API_KEY")},
)
if err != nil {
    log.Fatal(err)
}
resp, err := client.Spans.List(ctx, spans.ListRequest{
    Project: os.Getenv("ARIZE_PROJECT_NAME"),
    Space:   os.Getenv("ARIZE_SPACE_ID"),
    Limit:   5,
})
if err != nil {
    log.Fatal(err)
}
fmt.Printf("%d span(s) found\n", len(resp.Spans))

SDK span references: Python · TypeScript · Go.

Troubleshooting

No traces in Arize AX. Confirm ARIZE_SPACE_ID and ARIZE_API_KEY are set in the same shell that runs example.py. Enable OpenTelemetry debug logs with export OTEL_LOG_LEVEL=debug and re-run.
Google GenAI spans missing but other spans present. GoogleGenAIInstrumentor().instrument(...) must run before any from google import genai import. Make sure instrumentation.py is the first import in your entry point.
401 / 403 from Gemini. Verify GEMINI_API_KEY is set and has access to the model in the example. Swap gemini-3.8-flash for a model your key can call.
404 NOT_FOUND for the model. Google occasionally retires older Gemini aliases for new users. If gemini-3.8-flash returns 404, list models with client.models.list() and pick a current one.
Using Vertex AI instead of the Gemini API. Configure the SDK against Vertex per Google’s GenAI SDK docs (set GOOGLE_GENAI_USE_VERTEXAI, GOOGLE_CLOUD_PROJECT, GOOGLE_CLOUD_LOCATION). The same instrumentor captures Vertex calls — only the credential setup differs.
Go: spans never arrive. Spans are batched and only sent when tp.Shutdown runs. Keep the defer that calls it, and don’t exit with os.Exit or log.Fatal, which skip deferred calls.
Go: the span has no prompt, response, or token counts. Only the attributes set in main.go are recorded. If you move the Gemini call into your own function, pass the ctx returned by Start and set the output and token attributes on the same span.

OpenTelemetry

LLM Providers

Agent Frameworks

Coding Agents

Platforms

Orchestration

Vector Databases

Evaluation

Prerequisites

Launch Arize AX

Install

Configure credentials

Setup tracing

Run Google GenAI

Expected output

Verify in Arize AX

Check from the skill, CLI, or SDK

Troubleshooting

Resources

Gemini API Documentation

OpenInference Google GenAI Instrumentor

Google Gen AI Python SDK

Google Gen AI Go SDK

OpenInference Go Semantic Conventions

​Prerequisites

​Launch Arize AX

​Install

​Configure credentials

​Setup tracing

​Run Google GenAI

​Expected output

​Verify in Arize AX

​Check from the skill, CLI, or SDK

​Troubleshooting

​Resources

Gemini API Documentation

OpenInference Google GenAI Instrumentor

Google Gen AI Python SDK

Google Gen AI Go SDK

OpenInference Go Semantic Conventions

Prerequisites

Launch Arize AX

Install

Configure credentials

Setup tracing

Run Google GenAI

Expected output

Verify in Arize AX

Check from the skill, CLI, or SDK

Troubleshooting

Resources