Short answer
GPT-6.2 hasn't been released. As of November 4, 2026, OpenAI's model docs list GPT-6 Astra (gpt-6-astra) and no GPT-6.2 model, and OpenAI hasn't published a model ID. Two names have leaked but are not confirmed by OpenAI: gpt-6.2-astra showed up in some users' Codex usage breakdowns, and gpt-6.2-betelgeuse in a screenshot of an internal tracker shared on X. Background: GPT-6.2 Astra: what we know.
When it ships and is available on TeamoRouter, switching is small. In Codex, the OpenAI SDK and Claude Code, you change only the model name; the base URL and the API key stay the same. If you set up with GPT-6 Astra now, release day is a one-line change.
New accounts start with a $0 balance — top up before your first request. Top up from $5. If a request returns 402 (400 on the OpenAI format) with "wallet balance is insufficient", your balance is empty — your base URL and key are fine. Top up and retry.
Set up with GPT-6 Astra now; the same key is all you need on release day
Step 1: Confirm it's really out
Leaks and screenshots are not a model ID. On release day, check two places before you change any config:
- OpenAI's model docs. The ID listed there is the one to use. Don't copy an ID from a screenshot or a post on X.
- TeamoRouter's models page. It lists every model you can call with your key. We will also update the GPT-6.2 article with whether the model is live on TeamoRouter.
Reports say GPT-6.2 may come in more than one version, such as Astra and Sol, so there may be several new model IDs. Use the ones listed in OpenAI's model docs and pick by task, as you would between GPT-6 Astra and GPT-6.1 Sol today.
If you call an ID that isn't live yet, the request fails with an error instead of reaching a model. Switch back to gpt-6-astra and try again later.
Step 2: Codex
Codex keeps the endpoint in ~/.codex/config.toml; it doesn't read OPENAI_BASE_URL. Here is the setup with GPT-6 Astra, with your TeamoRouter key in OPENAI_API_KEY:
model_provider = "teamorouter"
model = "gpt-6-astra"
[model_providers.teamorouter]
name = "TeamoRouter"
base_url = "https://api.teamorouter.com/v1"
env_key = "OPENAI_API_KEY"
wire_api = "responses"
On release day you have two options:
- Try it in one session without touching the file:
codex -m <new model ID>. Your default staysgpt-6-astra. - Make it the default: change only the
model = "..."line. The[model_providers.teamorouter]block stays as it is.
Full Codex, Cursor and Claude Code setup: GPT-6 Astra in Codex, Cursor and Claude Code.
Step 3: The OpenAI SDK
Read the model name from an environment variable instead of writing it into the code. On release day you change one variable and redeploy nothing. Python:
import os
from openai import OpenAI
client = OpenAI(
base_url="https://api.teamorouter.com/v1",
api_key=os.environ["TEAMOROUTER_API_KEY"],
)
MODEL = os.environ.get("GPT_MODEL", "gpt-6-astra")
resp = client.responses.create(model=MODEL, input="Review this diff and list the risky changes: ...")
print(resp.output_text)
JavaScript:
import OpenAI from "openai";
const client = new OpenAI({
baseURL: "https://api.teamorouter.com/v1",
apiKey: process.env.TEAMOROUTER_API_KEY,
});
const model = process.env.GPT_MODEL ?? "gpt-6-astra";
const resp = await client.responses.create({ model, input: "Review this diff and list the risky changes: ..." });
console.log(resp.output_text);
If you want the new model when it's available and GPT-6 Astra when it isn't, catch the error and retry once with the fallback:
import openai
def ask(prompt, model, fallback="gpt-6-astra"):
try:
return client.responses.create(model=model, input=prompt)
except openai.APIStatusError as err:
if err.status_code in (400, 404) and model != fallback:
return client.responses.create(model=fallback, input=prompt)
raise
Chat Completions works the same way: client.chat.completions.create(model=MODEL, messages=[...]), with only the model name changing.
Step 4: Claude Code
Claude Code reaches GPT models through TeamoRouter's Anthropic-format endpoint. Point it at TeamoRouter once:
export ANTHROPIC_BASE_URL="https://api.teamorouter.com"
export ANTHROPIC_AUTH_TOKEN="sk-teamo-your-key"
Then pick the model:
- One session:
claude --model gpt-6-astra. On release day, use the new ID instead. - Every session:
export ANTHROPIC_MODEL="gpt-6-astra"and change the value later.
Claude Code shows a notice that a GPT model isn't in its built-in catalog and keeps the session within 200K tokens of context. That's expected; requests still go to the model you named.
Step 5: Try one real task before moving everything
- Price: OpenAI hasn't published GPT-6.2 pricing. For reference, OpenAI's price for GPT-6 Astra is $10 per million input tokens, $1.00 cached input and $50 output. Check the pricing page on release day before you move long agent sessions over.
- One task first: run the same refactor or bug fix you ran on GPT-6 Astra, and compare the result and the cost in your usage records.
- Keep the fallback: leave
gpt-6-astrain your configs as the default until the new model has handled a few real tasks.
What isn't known yet
The name (GPT-6.2 Astra, Betelgeuse or something else), the model ID, the price, the context window, and whether it reaches the API on day one or appears in ChatGPT and Codex first. We aren't guessing any of these. If it launches only inside OpenAI's own apps at first, the API ID won't work yet, so keep using gpt-6-astra.
Paying for usage
No subscription: you top up a balance from $5 and pay only for the tokens you use, by card or with USDT, USDC or ETH from any exchange app or wallet (how to pay with USDT). The same key also works for Claude, Gemini and DeepSeek.
FAQ
What's the GPT-6.2 model ID?
OpenAI hasn't published one. gpt-6.2-astra (seen in Codex usage breakdowns) and gpt-6.2-betelgeuse (seen in a leaked tracker screenshot) are unconfirmed. Use the ID in OpenAI's model docs once it's there.
Do I need a new key for GPT-6.2?
No. If GPT-6.2 is available on TeamoRouter, the same key and base URL work. Only the model name changes.
Will my Codex config break when GPT-6.2 comes out?
No. Your provider block doesn't change, and Codex keeps using the model you set until you change it.
Can I use GPT-6.2 on TeamoRouter today?
No, because OpenAI hasn't released it. GPT-6 Astra (gpt-6-astra), GPT-6.1 Sol and GPT-6.1 Terra are available now. See the GPT-6 Astra model page.