# OpenaiBatchAPI: A Python Library for OpenAI Batch API
[OpenAI Batch API Documentation](https://platform.openai.com/docs/guides/batch)
**Note:** *Currently supports only `gpt-4o` and `gpt-4o-mini` models.*
## Installation
Install the package from PyPI using [pip](http://www.pip-installer.org):
```bash
$ pip install OpenaiBatchAPI
```
## Usage Example
### Basic Example
```python
from openai_batch_api import OpenaiBatchAPI
# Initialize the Batch API client
batch_client = OpenaiBatchAPI(
api_key="API_TOKEN",
max_retry=3,
timeout=5 * 60 # Timeout in seconds
)
messages = []
# Add messages with custom IDs (e.g., "calc_{i}")
for i in range(7):
messages.append({
"id": f"calc_{i}",
"content": [
{
"role": "user",
"content": f"Calculate: 1 + {i} = "
}
]
})
# Add messages with auto-generated IDs (index-based)
for i in range(7):
messages.append({
"role": "user",
"content": f"Calculate: 1 + {i} = "
})
# Execute batch completion
batchs = batch_client.batchs_completion(
messages,
max_completion_tokens=32,
model="gpt-4o-mini",
temperature=0,
seed=42,
batch_size=2
)
# Print batch outputs
for batch in batchs:
print("BATCH ID:", batch.id)
print("FILE ID:", batch.output_file.id)
for content in batch.output_file.contents:
print("Message ID:", content["id"])
print("Response:", content["choices"][0]['message']['content'])
print()
```
### Output Sample
The script will output results similar to this:
```plaintext
Preparing: 100%|███████████████████████| 4/4 [00:10<00:00, 2.74s/it]
Sending: 100%|█████████████████████████| 4/4 [00:02<00:00, 1.60it/s]
Processing: 100%|██████████████████████| 4/4 [00:12<00:00, 3.10s/it]
BATCH ID: batch_672b237c62f88190b0d69cbf789eb754
FILE ID: file-IlJUOenuE8p7okKB3TX2ANKW
Message ID: calc_0
Response: 1 + 0 = 1
Message ID: calc_1
Response: 1 + 1 = 2.
BATCH ID: batch_672b237ce4fc8190813e7b1239f0336d
FILE ID: file-hTFNk0uBC37plDaWtgYtD0Dr
Message ID: calc_2
Response: 1 + 2 = 3.
Message ID: calc_3
Response: 1 + 3 = 4.
BATCH ID: batch_672b237d937c8190a2ec942aa9401732
FILE ID: file-AbzBSJGl2pPU1B6KPmMuIDBB
Message ID: calc_4
Response: 1 + 4 = 5.
Message ID: calc_5
Response: 1 + 5 = 6.
BATCH ID: batch_672b237e076c8190807819fde4cdd093
FILE ID: file-jRAePmaheNSNAWYcawdmSgYF
Message ID: calc_6
Response: 1 + 6 = 7.
```
### Usage Statistics
The `print_usage_table` method outputs a table summarizing token usage and cost.
```python
batch_client.print_usage_table(digits=8)
```
This outputs:
```plaintext
+------------+--------+-------------+
| Category | Tokens | Price |
+------------+--------+-------------+
| Prompt | 119 | $0.00000892 |
| Completion | 55 | $0.00001650 |
+------------+--------+-------------+
| Total | 174 | $0.00002543 |
+------------+--------+-------------+
```
Raw data
{
"_id": null,
"home_page": "https://github.com/it-dainb/batch_api.git",
"name": "OpenaiBatchAPI",
"maintainer": null,
"docs_url": null,
"requires_python": "<=3.12,>=3.8",
"maintainer_email": null,
"keywords": "openai batch_api batch api",
"author": "IT.DAINB",
"author_email": "it.dainb@gmail.com",
"download_url": null,
"platform": null,
"description": "# OpenaiBatchAPI: A Python Library for OpenAI Batch API\n[OpenAI Batch API Documentation](https://platform.openai.com/docs/guides/batch)\n\n**Note:** *Currently supports only `gpt-4o` and `gpt-4o-mini` models.*\n\n## Installation\n\nInstall the package from PyPI using [pip](http://www.pip-installer.org):\n\n```bash\n$ pip install OpenaiBatchAPI\n```\n\n## Usage Example\n\n### Basic Example\n\n```python\nfrom openai_batch_api import OpenaiBatchAPI\n\n# Initialize the Batch API client\nbatch_client = OpenaiBatchAPI(\n api_key=\"API_TOKEN\",\n max_retry=3,\n timeout=5 * 60 # Timeout in seconds\n)\n\nmessages = []\n\n# Add messages with custom IDs (e.g., \"calc_{i}\")\nfor i in range(7):\n messages.append({\n \"id\": f\"calc_{i}\",\n \"content\": [\n {\n \"role\": \"user\",\n \"content\": f\"Calculate: 1 + {i} = \"\n }\n ]\n })\n\n# Add messages with auto-generated IDs (index-based)\nfor i in range(7):\n messages.append({\n \"role\": \"user\",\n \"content\": f\"Calculate: 1 + {i} = \"\n })\n\n# Execute batch completion\nbatchs = batch_client.batchs_completion(\n messages,\n max_completion_tokens=32,\n model=\"gpt-4o-mini\",\n temperature=0,\n seed=42,\n batch_size=2\n)\n\n# Print batch outputs\nfor batch in batchs:\n print(\"BATCH ID:\", batch.id)\n print(\"FILE ID:\", batch.output_file.id)\n for content in batch.output_file.contents:\n print(\"Message ID:\", content[\"id\"])\n print(\"Response:\", content[\"choices\"][0]['message']['content'])\n print()\n```\n\n### Output Sample\n\nThe script will output results similar to this:\n\n```plaintext\nPreparing: 100%|\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588| 4/4 [00:10<00:00, 2.74s/it]\nSending: 100%|\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588| 4/4 [00:02<00:00, 1.60it/s]\nProcessing: 100%|\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588| 4/4 [00:12<00:00, 3.10s/it]\n\nBATCH ID: batch_672b237c62f88190b0d69cbf789eb754\nFILE ID: file-IlJUOenuE8p7okKB3TX2ANKW\nMessage ID: calc_0\nResponse: 1 + 0 = 1\nMessage ID: calc_1\nResponse: 1 + 1 = 2.\n\nBATCH ID: batch_672b237ce4fc8190813e7b1239f0336d\nFILE ID: file-hTFNk0uBC37plDaWtgYtD0Dr\nMessage ID: calc_2\nResponse: 1 + 2 = 3.\nMessage ID: calc_3\nResponse: 1 + 3 = 4.\n\nBATCH ID: batch_672b237d937c8190a2ec942aa9401732\nFILE ID: file-AbzBSJGl2pPU1B6KPmMuIDBB\nMessage ID: calc_4\nResponse: 1 + 4 = 5.\nMessage ID: calc_5\nResponse: 1 + 5 = 6.\n\nBATCH ID: batch_672b237e076c8190807819fde4cdd093\nFILE ID: file-jRAePmaheNSNAWYcawdmSgYF\nMessage ID: calc_6\nResponse: 1 + 6 = 7.\n```\n\n### Usage Statistics\n\nThe `print_usage_table` method outputs a table summarizing token usage and cost.\n\n```python\nbatch_client.print_usage_table(digits=8)\n```\n\nThis outputs:\n\n```plaintext\n+------------+--------+-------------+\n| Category | Tokens | Price |\n+------------+--------+-------------+\n| Prompt | 119 | $0.00000892 |\n| Completion | 55 | $0.00001650 |\n+------------+--------+-------------+\n| Total | 174 | $0.00002543 |\n+------------+--------+-------------+\n```\n",
"bugtrack_url": null,
"license": "Apache License 2.0",
"summary": "OpenaiBatchAPI: A Python Library that supports OpenAI Batch API",
"version": "1.1.2",
"project_urls": {
"Homepage": "https://github.com/it-dainb/batch_api.git"
},
"split_keywords": [
"openai",
"batch_api",
"batch",
"api"
],
"urls": [
{
"comment_text": "",
"digests": {
"blake2b_256": "121d12e5a9b8736d524454e221a079a16bdbd6668a4276638643d95da40f9184",
"md5": "88104a4b4911db7f28aaee6b00f20181",
"sha256": "d0a5bf6e852b4adfdbab8033c1b94fd2fb67ec47c31231a32470c1ddcfbb621e"
},
"downloads": -1,
"filename": "OpenaiBatchAPI-1.1.2-py2.py3-none-any.whl",
"has_sig": false,
"md5_digest": "88104a4b4911db7f28aaee6b00f20181",
"packagetype": "bdist_wheel",
"python_version": "py2.py3",
"requires_python": "<=3.12,>=3.8",
"size": 7222,
"upload_time": "2024-11-07T10:39:52",
"upload_time_iso_8601": "2024-11-07T10:39:52.974692Z",
"url": "https://files.pythonhosted.org/packages/12/1d/12e5a9b8736d524454e221a079a16bdbd6668a4276638643d95da40f9184/OpenaiBatchAPI-1.1.2-py2.py3-none-any.whl",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2024-11-07 10:39:52",
"github": true,
"gitlab": false,
"bitbucket": false,
"codeberg": false,
"github_user": "it-dainb",
"github_project": "batch_api",
"travis_ci": false,
"coveralls": false,
"github_actions": true,
"requirements": [],
"lcname": "openaibatchapi"
}
JSON
