Skip to content

The model you choose is the model you get.

Routed to the model you choose — no downgrades, no swaps.

One API key for multiple AI models. Priced in USD, topped up in CNY.

Get an API key View pricing

Base URL

OpenAI-compatible
https://api.6090ti.com/v1
Anthropic-compatible
https://api.6090ti.com
  • No downgrades, no swaps

    Your request goes to the model you called.

  • Pay as you go

    No monthly fee. You pay per call.

  • Usage on record

    Recorded usage and charges, always in your console.

  • Self-service

    Keys, balance and orders — all in one console.

Model pricing

Priced in USD.

ModelInput (official)/ 1M tokensOutput (official)/ 1M tokensInput (6090ti)/ 1M tokensOutput (6090ti)/ 1M tokens
claude-fable-5-2$8.00Cached input $0.20$40.00$1.68 79% offCached input $0.042$8.40 79% off
claude-fable-5-1$10.00Cached input $0.25$50.00$2.10 79% offCached input $0.0525$10.50 79% off
claude-fable-5$10.00Cached input $1.00$50.00$2.10 79% offCached input $0.21$10.50 79% off
claude-opus-5-5$4.00Cached input $0.20$20.00$0.72 82% offCached input $0.036$3.60 82% off
claude-opus-5$5.00Cached input $0.50$25.00$0.90 82% offCached input $0.09$4.50 82% off
claude-sonnet-5-5$2.00Cached input $0.10$10.00$0.36 82% offCached input $0.018$1.80 82% off
claude-haiku-5-5$0.50Cached input $0.05$2.50$0.115 77% offCached input $0.0115$0.575 77% off

Full pricing table →

Quick start

Point your tool or code at 6090ti and enter your API key plus a model name. Replace the parts in angle brackets with your own values; your API key goes where it is highlighted in green.

Claude Code

export ANTHROPIC_BASE_URL="https://api.6090ti.com"
export ANTHROPIC_API_KEY="<your key>"
export ANTHROPIC_MODEL="<model>"
claude

Anthropic-compatible format. Put it in the env block of ~/.claude/settings.json, or set it temporarily like this and run claude.

Codex

model = "<model>"
model_provider = "p6090ti"

[model_providers.p6090ti]
name = "6090ti"
base_url = "https://api.6090ti.com/v1"
env_key = "P6090TI_API_KEY"
wire_api = "responses"

OpenAI-compatible format (Responses). Put this in ~/.codex/config.toml and the key in the environment variable P6090TI_API_KEY.

OpenCode

{
  "$schema": "https://opencode.ai/config.json",
  "provider": {
    "p6090ti": {
      "npm": "@ai-sdk/openai-compatible",
      "name": "6090ti",
      "options": { "baseURL": "https://api.6090ti.com/v1" },
      "models": { "<model>": { "name": "<model>" } }
    }
  }
}

OpenAI-compatible format. Put this in opencode.json in your project, then save the key with /connect.

Python

from openai import OpenAI

client = OpenAI(base_url="https://api.6090ti.com/v1", api_key="<your key>")
resp = client.chat.completions.create(
    model="<model>",
    messages=[{"role": "user", "content": "Hello"}],
)
print(resp.choices[0].message.content)

OpenAI-compatible format, with the OpenAI SDK.

curl

curl https://api.6090ti.com/v1/models \
  -H "Authorization: Bearer <your key>"

List the models your key can use.

Full guides →

Set up your tools with our guides

OpenAI- and Anthropic-compatible formats are supported, so one key works in the tools you already use.

All guides →

Latest articles

All articles →

FAQ

Will my model be swapped for a cheaper one?
No. Requests go to the model you choose.
How is usage billed?
Each request is charged from your balance under the pricing rules for the selected model and request. Prices are in USD; top-ups are paid in CNY.
What if my API key leaks?
Disable the leaked key right away, then create a new one and check for unusual usage. Changing your password does not revoke existing keys.
What is the base URL?
OpenAI-compatible format: https://api.6090ti.com/v1; Anthropic-compatible format: https://api.6090ti.com.
Do I need a separate key for each tool?
No. One key calls every model enabled for you; you can still create one per computer or project to see usage separately.
How do I see which models a key can use?
Request /v1/models with that key; the response lists the available models.

More FAQ →