BedrockのOpenAIモデル（GPT-5.5 / GPT-5.4）がus-east-1で利用可能になったので試してみた

from openai import OpenAI
from aws_bedrock_token_generator import BedrockTokenGenerator

token_generator = BedrockTokenGenerator(region="us-east-1")
client = OpenAI(
    base_url="https://bedrock-mantle.us-east-1.api.aws/openai/v1",
    api_key=token_generator(),
)

response = client.responses.create(
    model="openai.gpt-5.5",
    input="Explain Amazon Bedrock in one sentence.",
    reasoning={"effort": "low"},
)
print(response.output_text)
print(f"Usage: input={response.usage.input_tokens}, output={response.usage.output_tokens}")

import time
from openai import OpenAI
from aws_bedrock_token_generator import BedrockTokenGenerator

token_generator = BedrockTokenGenerator(region="us-east-1")
client = OpenAI(
    base_url="https://bedrock-mantle.us-east-1.api.aws/openai/v1",
    api_key=token_generator(),
)

start = time.time()
first_token_time = None
full_text = ""

stream = client.responses.create(
    model="openai.gpt-5.5",
    input="Amazon Bedrock の主な特徴を3つ挙げてください。",
    reasoning={"effort": "low"},
    stream=True,
)

for event in stream:
    if event.type == "response.output_text.delta":
        if first_token_time is None:
            first_token_time = time.time()
        full_text += event.delta
    elif event.type == "response.completed":
        usage = event.response.usage

elapsed = time.time() - start
ttft = first_token_time - start if first_token_time else None

print(f"TTFT: {ttft:.2f}s")
print(f"Total: {elapsed:.2f}s")
print(f"Usage: input={usage.input_tokens}, output={usage.output_tokens}")
print(full_text)

response.created
→ response.in_progress
→ response.output_item.added
→ response.content_part.added
→ response.output_text.delta（N回繰り返し）
→ response.output_text.done
→ response.content_part.done
→ response.output_item.done
→ response.completed

# これは動作しない
response = client.chat.completions.create(
    model="openai.gpt-5.5",
    messages=[{"role": "user", "content": "Hello"}],
    max_tokens=256,
)

openai.BadRequestError: Error code: 400 - {
  "error": {
    "code": "validation_error",
    "message": "The model 'openai.gpt-5.5' does not support the '/v1/chat/completions' API",
    "param": null,
    "type": "invalid_request_error"
  }
}