Every major model.
Flux picks the right one for each job, or you pin your favourite.
What your agent can do with it
- Send hard planning and debugging to the Reasoning lane and quick edits to flux-fast, or let flux-auto pick among vetted models for each kind of task, cost-aware.
- Pin one exact model when you want the same one every time.
- Try a different model without a new account, key or SDK.
- See which model answered each request, so nothing is a black box.
How it reaches your agent
Swap the base URLLive
Point your agent’s OpenAI or Anthropic base URL at Flux and use your Flux key. Set the model to flux-auto and Flux picks among vetted models for each kind of task, cost-aware.
Pick a lane or pin a modelLive
Use flux-fast, flux-standard or flux-reasoning to choose a lane, or a pinned model name to skip the picker.
The Flux MCP serverLive
MCP clients get routing tools on the same key: route a request, explain a pick and list the models.
Works with Murage, Wayland, Claude Code, Codex, Cursor, Hermes, OpenClaw, Cline, and anything else with an OpenAI or Anthropic endpoint. See the setup guides.
Price
- Autopilot (flux-auto)$2 in / $8 out per 1M tokensThe Standard rate, or the Fast rate when flux-auto sends your request to the Fast lane and a Fast model answers. Never more than Standard.
- Reasoning (flux-reasoning)$4 in / $15 out per 1M tokens
- Standard (flux-standard)$2 in / $8 out per 1M tokens
- Fast (flux-fast)$1 in / $4 out per 1M tokens
- A pinned modelPer token, at that model’s own rate. How pinning works
These are plan prices. Pay as you go adds 12%, applied once.
Limits
- Pinning a model skips the picker, and you pay that model’s rate.
- flux-auto bills at the Standard rate, or the Fast rate when it sends your request to the Fast lane and a Fast model answers. Never more than Standard.
Why not wire it up yourself?
One account instead of several
No separate sign-up, key and billing page for each provider behind an ability.
One key, one bill
Every ability is billed to your Flux account, on one bill.
Nothing new to maintain
No extra SDKs to study or keep current. Your agent talks to one endpoint it already knows.
Questions
Which model answers when I use flux-auto?
Flux picks among vetted models for each kind of task, cost-aware. Each response tells you which model answered.
What does flux-auto cost?
It bills at the Standard rate, or the Fast rate when it sends your request to the Fast lane and a Fast model answers. It never costs more than Standard, even when a hard prompt reaches the Reasoning lane. The rates are in the price block above and on the Pricing page.
Do I need a new SDK?
No. Flux speaks the OpenAI and Anthropic formats, so most agents only need a new base URL and key.
Can I always use the same model?
Yes. Pin it by name. Pinning skips the picker, and you pay that model’s own rate.
One key. Everything behind it.
Your agent keeps its brain. Flux gives it the rest, on one key and one bill.