How can I connect an internally deployed custom model API to the OpenAI agents SDK?

[keepwonder]

Please read this first

• Have you read the docs?Agents SDK docs yes
• Have you searched for related issues? Others may have had similar requests yes

Question

Describe your question. Provide details if available.

below is the common api call

import json, uuid, math, time, base64, hashlib, datetime, requests

url = "http://xxx.xx.xx.xx:xxx/chat/completions/inference-xxxx-xxxx"

appid = "xx-xx-xx"
appKey = "96e8dfghghjjhgjxcvwytium11gfgf87dsd3adffdsfsd" 
capabilityname = "000000000000000000000000"

_uuid = "".join(str(uuid.uuid4()).split('-'))

xServerParam = {
    "appid": appid,
    "csid": appid + capabilityname + _uuid
}
xCurTime = str(math.floor(time.time()))
xServerParam = str(base64.b64encode(json.dumps(xServerParam).encode('utf-8')), encoding="utf8")
xCheckSum = hashlib.md5(bytes(appKey + xCurTime + xServerParam, encoding="utf8")).hexdigest()
print(xServerParam)
headers = {
    "X-Server-Param": xServerParam,
    "X-CurTime": xCurTime,
    "X-CheckSum": xCheckSum,
    'content-type': 'application/json;charset=UTF-8',
    'Accept': 'text/event-stream'
}

query = "hello,tell me about yourself"

data = {
    "model": "Qwen2.5-7B-Instruct",
    "repetition_penalty": 1.05,
    "temperature": 0.8,
    "top_p": 0.9,
    "top_k": 20,
    "messages": [
        {"role": "system", "content": "You are a helpful assistant."},
        {"role": "user", "content": query}
    ],
    "stream": True
}

first = True
result = ""
print("[{}] send request".format(datetime.datetime.now().strftime('%Y-%m-%d %H:%M:%S:%f')))
start_time = time.time()
response = requests.post(url, headers=headers, data=json.dumps(data), stream=True)
print(response)

for line in response.iter_lines():
    if line:
        text = line.decode("utf-8")
        print(text, end="\n", flush=True)

end_time = time.time()
print("time cost: {:.4f}s".format(end_time - `start_time))`

response is like this , it's an openai format .

I try some way to integerated, but not worked!

I want to know is agents-python support these kind of work?

If support, how to achive？

[seratch]

When you use Chat Completions API compatible model, you can go with 1) OpenAIChatCompletionsModel or 2) https://openai.github.io/openai-agents-python/models/litellm/ Most Chat Completions API models are not 100% compatible, so relying on LiteLLM's adapter layer should be more stable.

[github-actions]

This issue is stale because it has been open for 7 days with no activity.

[github-actions]

This issue was closed because it has been inactive for 3 days since being marked as stale.