What does Codex cost? And how to pay 20 to 25% less per token.
Codex comes with ChatGPT Plus at $20 a month and Pro from $100, or runs on an API key billed per token. Point it at APIsRouter and GPT-6 Sol costs $1.50 / $8 per million input and output tokens instead of $2 / $10.
Example: a month of Codex on GPT-6 Sol, per token
Example task: 100,000 input tokens (90,000 cached) and 5,000 output; real tasks vary. Plus costs $20 a month, Pro from $100.
| OpenAI API | APIsRouter | |
|---|---|---|
| Weekends, 50 tasks | $4.40 | $3.43 |
| Every workday, 400 tasks | $35.20 | $27.40 |
| All day, 2,000 tasks | $176 | $137 |
Every way to pay for Codex
A plan trades a fixed price for usage limits. An API key has no 5-hour or weekly windows and bills each token.
| Option | Price | How usage works | Suits |
|---|---|---|---|
| ChatGPT Free | $0 | GPT-6 Luna in the Codex desktop app, where it has rolled out | Trying Codex on small tasks |
| ChatGPT Go | $8 a month | GPT-6 Luna in the Codex desktop app | Light tasks now and then |
| ChatGPT Plus | $20 a month | About 15 to 160 local messages per 5 hours on GPT-6.1 Sol; weekly limits may also apply | A few focused sessions a week |
| ChatGPT Pro | $100, $200 or $500 a month | No 5-hour limit for now | Coding all day, every day |
| OpenAI API key | Per token, at OpenAI's list | No 5-hour or weekly windows. CLI, SDK and IDE only, no cloud tasks | Automation and CI |
| APIsRouter | Per token, 20 to 25% under OpenAI's list on GPT-6 Sol and GPT-6 Astra | No 5-hour or weekly windows, no subscription. CLI, SDK and IDE only. Credit never expires | Paying only for the weeks you code |
Codex model prices, per million tokens
OpenAI's API list for the GPT models Codex runs. Where APIsRouter sells a model below OpenAI, both prices are shown. OpenAI prices from developers.openai.com/api/docs/pricing, standard tier, September 2026.
| Model | Input: OpenAI / APIsRouter | Cached input: OpenAI / APIsRouter | Output: OpenAI / APIsRouter |
|---|---|---|---|
| GPT-6.1 Sol (gpt-6.1-sol) | $2 / $1.50 | $0.10 / $0.08 | $10 / $8 |
| GPT-6 Sol (gpt-6-sol) | $2 / $1.50 | $0.20 / $0.15 | $10 / $8 |
| GPT-6 Astra (gpt-6-astra) | $10 / $8 | $1 / $0.80 | $50 / $40 |
| GPT-6 Luna (gpt-6-luna) | $0.10 / $0.08 | $0.01 / $0.008 | $0.50 / $0.40 |
Run Codex on APIsRouter
- Get a key: Start free on the $2 trial credit, or top up from $10 below. A top-up key arrives by email.
- Add APIsRouter to Codex: Add these lines at the top of ~/.codex/config.toml, replacing any model lines already there.
- Run codex: Export APISROUTER_API_KEY with your key, then run codex as usual.
model = "gpt-6-sol"
model_provider = "apisrouter"
[model_providers.apisrouter]
name = "APIsRouter"
base_url = "https://api.apisrouter.com/v1"
env_key = "APISROUTER_API_KEY"
wire_api = "responses"Two ways to pay
Both fill one balance, spent at the prices above.
FAQ
How much does Codex cost?
Codex comes with ChatGPT: free with GPT-6 Luna where it has rolled out, $8 a month on Go, $20 on Plus and $100, $200 or $500 on Pro. On an API key you pay per token instead: GPT-6 Sol costs $2 / $10 per million input and output tokens at OpenAI and $1.50 / $8 on APIsRouter.
Is Codex free?
The Free plan includes Codex with GPT-6 Luna in the desktop app, where it has rolled out. To run GPT-6 Sol without a subscription, start on APIsRouter's $2 trial credit: sign in and claim it from your dashboard, no card needed.
Is ChatGPT Plus enough for Codex?
Plus allows about 15 to 160 local messages per 5 hours on GPT-6.1 Sol, and weekly limits may also apply. If you hit them, buy ChatGPT credits, move to Pro, or pay per token with an API key.
Can I use Codex with an API key?
Yes. Codex runs on an API key in the CLI, the SDK and the IDE extension; cloud features such as GitHub code review and the Slack integration need a ChatGPT plan. The config above points Codex at APIsRouter.
Which model should I use in Codex through APIsRouter?
GPT-6 Sol for everyday coding at $1.50 / $8 per million input and output tokens, and GPT-6 Astra for the hardest tasks at $8 / $40.
What happens when my balance runs out?
Requests stop until you top up. There are no overage bills, and unused credit never expires.