selected model is gpt5 but ai response gpt 4..

[shell jay]

from openai import OpenAI
import os
from dotenv import load_dotenv
load_dotenv()

client = OpenAI(
  base_url="https://openrouter.ai/api/v1",
  api_key=os.getenv("OPENROUTER_API_KEY"),
)

completion = client.chat.completions.create(

  extra_body={},
  model="openai/gpt-5-chat",
  messages=[
    {
      "role": "user",
      "content": [
        {
          "type": "text",
          "text": "What is ur model ?"
        },

      ]
    }
  ]
)
print(completion.choices[0].message.content)

I’m based on GPT‑4, OpenAI’s large language model. I can understand and generate text, interpret images, and help with reasoning, explanation, problem-solving, and creativity.
this is response i get.. why ?

[obsidian rain]

I'm an AI assistant - my answers may be incorrect and I cannot see screenshots. Please verify information with the OpenRouter docs.

Here are possible reasons why you're getting a GPT-4-based response even though you selected "openai/gpt-5-chat":

1. Availability and Routing:
   The OpenRouter Models page states that some models, like GPT-5, may require special configuration (such as BYOK) or may not always be available. If "openai/gpt-5-chat" is unavailable or not properly configured for your account, OpenRouter might automatically route your request to a fallback model—often GPT-4 or GPT-4 Turbo—without an explicit error. This can result in responses that identify as GPT-4, even if you requested GPT-5.

2. Model Self-Disclosure:
   Many models, including experimental or preview models, may still self-identify as "GPT-4" if their internal system prompts have not been updated, even when a newer model is used. This is common for preview or unreleased models.

3. BYOK Requirement:
   According to the GPT-5 model description, "*Note that BYOK is required for this model. Set up here: https://openrouter.ai/settings/integrations*." If you haven't set up BYOK (Bring Your Own Key), requests to "openai/gpt-5-chat" will not actually use GPT-5 and may silently fall back to another available model.

What to do:

• Double-check that your account has access to "openai/gpt-5-chat" and that BYOK is set up if required.
• You can search for the exact model and its requirements at OpenRouter Models.
• If you believe you should have access but still get fallback responses, contact mods through the Discord server for help.

References:

• "*GPT-5 i...

...s OpenAI’s most advanced model...Note that BYOK is required for this model. Set up here: https://openrouter.ai/settings/integrations*" GPT-5 Model

• See available models and requirements: All OpenRouter Models

If you need further assistance, please provide:

• The full API response JSON (including headers)
• Confirmation if you have BYOK configured
• Any error messages or logs you receive

This will help mods diagnose the issue more precisely.

-# This is an automated response powered by Answer HQ. Use !mute to stop the bot from responding.

[shell jay]

from openai import OpenAI
import os
from dotenv import load_dotenv
load_dotenv()

client = OpenAI(
  base_url="https://openrouter.ai/api/v1",
  api_key=os.getenv("OPENROUTER_API_KEY"),
)

completion = client.chat.completions.create(

  extra_body={},
  model="openai/gpt-4o-mini",
  messages=[
    {
      "role": "user",
      "content": [
        {
          "type": "text",
          "text": "What is ur model ?"
        },

      ]
    }
  ]
)
print(completion.choices[0].message.content)

I am based on OpenAI's GPT-3 model, which is a state-of-the-art language processing AI designed to understand and generate human-like text. If you have any specific questions about capabilities or features, feel free to ask!
it response like this

requested gpt 4o mini but it reacts gpt 3 model @obsidian rain

[snow cobalt]

the bot is wrong, the models dont know what model they are because the name is usually picked only after training is done therefore the model never learnt that it is GPT 5

the reason it knows what model it is for example in ChatGPT is because its told in the sytem prompt

[urban nest]

That's not true. I tested openai/gpt-5-chat using openrouter's UI and using the api. In the UI the model answer was always GPT-5, and in the API aways GPT-4

[snow cobalt]

because in the ui settings it injects a prompt telling it what model it is.

[urban nest]

Very good point, you're absolutely right

[shell jay]

how to disable default system prompt in ui.. @snow cobalt

[snow cobalt]

click on the triple dots next to the model and it should be above the system prompt thing

[shell jay]

thank u so much..one more question what is the use of presets @snow cobalt

[snow cobalt]

you can simply make presets for certain settings as one "model"

look at the docs im not very familiar

with them

[shell jay]

that is for chat ui or api ?

[snow cobalt]

should work for both i believe