Implementation:Googleapis Python genai Batches

Knowledge Sources	Googleapis_Python_genai Google GenAI SDK Docs
Domains	Batch_Processing, Generative_AI
Last Updated	2026-02-15 14:00 GMT

Overview

Concrete tool for batch prediction processing provided by the Google Gen AI SDK.

Description

The Batches and AsyncBatches classes provide API modules for creating, managing, and monitoring batch prediction jobs. They support text generation, embedding, and multi-modal batch processing with sources from GCS, BigQuery, or inlined requests. Both Gemini Developer API and Vertex AI backends are supported.

Usage

Import these classes when you need to process large volumes of content generation or embedding requests as batch jobs rather than individual API calls, enabling cost-efficient offline processing.

Code Reference

Source Location

• Repository: Googleapis_Python_genai
• File: google/genai/batches.py
• Lines: 1504-2600 (Batches at L1504-2046, AsyncBatches at L2049-2600)

Signature

class Batches(_api_module.BaseModule):
    def create(
        self,
        *,
        model: str,
        src: types.BatchJobSourceUnionDict,
        config: Optional[types.CreateBatchJobConfigOrDict] = None,
    ) -> types.BatchJob: ...

    def get(self, *, name: str, config: Optional[types.GetBatchJobConfigOrDict] = None) -> types.BatchJob: ...
    def cancel(self, *, name: str, config: Optional[types.CancelBatchJobConfigOrDict] = None) -> None: ...
    def delete(self, *, name: str, config: Optional[types.DeleteBatchJobConfigOrDict] = None) -> types.DeleteResourceJob: ...
    def list(self, *, config: Optional[types.ListBatchJobsConfigOrDict] = None) -> Pager[types.BatchJob]: ...

    def create_embeddings(
        self,
        *,
        model: str,
        src: types.EmbeddingsBatchJobSourceOrDict,
        config: Optional[types.CreateEmbeddingsBatchJobConfigOrDict] = None,
    ) -> types.BatchJob: ...

Import

from google import genai
client = genai.Client(api_key="...")
# Access via: client.batches

I/O Contract

Inputs

Name	Type	Required	Description
model	str	Yes	Model ID for batch processing (e.g. "gemini-2.0-flash")
src	BatchJobSourceUnionDict	Yes	Data source (GCS URI, BigQuery table, inlined requests, or file name)
config	CreateBatchJobConfigOrDict	No	Batch job configuration including destination
name	str	For get/cancel/delete	Batch job resource name

Outputs

Name	Type	Description
create() returns	BatchJob	Created batch job with name, state, metadata
get() returns	BatchJob	Batch job details and current state
list() returns	Pager[BatchJob]	Paginated list of batch jobs
delete() returns	DeleteResourceJob	Deletion operation result

Usage Examples

Create Batch Job from GCS

from google import genai
from google.genai import types

client = genai.Client(vertexai=True, project="my-project", location="us-central1")

# Create batch job from GCS source
batch_job = client.batches.create(
    model="gemini-2.0-flash",
    src="gs://my-bucket/input.jsonl",
    config=types.CreateBatchJobConfig(
        dest="gs://my-bucket/output/"
    ),
)

print(f"Batch job: {batch_job.name}, State: {batch_job.state}")

List and Monitor Batch Jobs

# List all batch jobs
for job in client.batches.list():
    print(f"{job.name}: {job.state}")

# Get specific batch job status
job = client.batches.get(name="batches/my-batch-job-id")
print(f"State: {job.state}")

Related Pages

• Environment:Googleapis_Python_genai_Python_3_10_SDK_Runtime