Request to enable GPT-3.5 Turbo 16K on my OpenRouter account

[crisp oracle]

Hi! Could you please enable the model openai/gpt-3.5-turbo-16k for my OpenRouter account?
I’d like to use it via API for a voice-response integration project. Thanks a lot!

[thin plaza]

but why lol

unless you're doing a research paper on how models from yesteryear would fair as voice agents.. there are better models in every way for cheaper
this is deprecated
try gpt-4.1-nano (lowest latency/highest throughput openai model) or gemini 2.0 flash

[crisp oracle]

thanks you

Hi again, thank you for your advice.
I tried using gpt-4.1-nano as recommended, but unfortunately it doesn’t activate either.
The message I keep receiving is: “The model 'account/usage' is not available”, even though I’ve added credits and set everything up (API key, endpoint, voice integration, etc.).
This seems unrelated to a specific model and might be due to a general issue with account permissions or access.
I’d really appreciate any help or clarification. Thank you so much in advance!

[thin plaza]

@warped depot

[warped depot]

can you share your request?

[crisp oracle]

response = requests.post(
"https://openrouter.ai/api/v1/chat/completions",
headers={
"Authorization": f"Bearer {OPENROUTER_API_KEY}",
"Content-Type": "application/json"
},
json={
"model": "openai/gpt-3.5-turbo-16k",
"messages": [
{"role": "system", "content": "You are Monica, a sensual AI voice."},
{"role": "user", "content": user_message}
]
}
)

We also tried with "openrouter/openai/gpt-4.1-nano" and "openrouter/openai/gpt-3.5-turbo" but none of them return a usable response.

We receive either a generic error or nothing at all — no "choices" in the response. Audio synthesis with ElevenLabs works fine separately.

Thanks a lot for helping us figure this out

[warped depot]

the model should be openai/gpt-4.1-nano

can you also post what the actual full response is

[crisp oracle]

Thanks! We had actually tested the 4.1-nano first, as you suggested, and it was the only one that returned a valid response. However, since our project requires longer context windows, we switched back to gpt-3.5-turbo-16k. That’s when we started getting errors or empty responses.

We’ll share the full response content shortly. Thank you again!

[warped depot]

the 4.1 model series has a much larger context window than the 3.5

3.5 is a very old model family and is definitely not recommended for production use

openai/gpt-3.5-turbo-16k has a 16,000 token context window.

openai/gpt-4.1-nano / openai/gpt-4.1-mini / openai/gpt-4.1 have a 1 million context window and are much newer than the now deprecated 3.5 model family

[crisp oracle]

ok i changed. but gives me that. it s now gpt 4.1 mini ,

[thin plaza]

how are you getting this error?

this happens if you go to URL https://openrouter.ai/accounts/usage

I am not sure you are doing your API requests right

you probably had AI generate some bad code?

hallucinating endpoints and such

first of all, confirm that the api request for gpt 4.1 nano is actually going through: https://openrouter.ai/activity