Read this first. Anthropic does not support routing Claude Code to non-Claude models through any gateway. Claude Code works with Smart and Fast through Uno, but tool calls can fail where they would not on Claude. Treat it as a way to keep working when your plan's limit runs out, and check the result. On 16 October 2026 we pointed Claude Code 2.1.284, Codex 0.154.0 and opencode 1.18.32 at Uno with a test key: each reached Uno's endpoint and got Uno's own answer to a wrong key. A full run with a real key and tool calls is still to do.
Steps
-
Copy your Uno key
Open the Uno console: Uno AI, then Use Uno AI from your code. Copy the API key. Keep it like a password: it spends your hours and your balance.
-
Test the key with one request
Before you change Claude Code, ask Uno one short question in Claude's own format. An answer with
"content"means the address and the key work.$ export UNO_API_KEY=your-uno-key $ curl -sS https://console.uno.place/v1/messages \ -H "Authorization: Bearer $UNO_API_KEY" \ -H "anthropic-version: 2023-06-01" \ -H "content-type: application/json" \ -d '{"model": "uno/fast", "max_tokens": 50, "messages": [{"role": "user", "content": "Say hi"}]}'
-
Point Claude Code at Uno
Put the settings in
~/.claude/settings.json: the env block of ~/.claude/settings.json applies everywhere Claude Code runs, and wins over a shell export. The address has no/v1at the end: Claude Code adds/v1/messagesitself.{ "env": { "ANTHROPIC_BASE_URL": "https://console.uno.place", "ANTHROPIC_AUTH_TOKEN": "your-uno-key", "ANTHROPIC_MODEL": "uno/smart", "ANTHROPIC_DEFAULT_HAIKU_MODEL": "uno/fast", "CLAUDE_CODE_SUBAGENT_MODEL": "uno/fast" } }Smart writes the code. Fast takes the background jobs and the subagents: ANTHROPIC_DEFAULT_HAIKU_MODEL sets what the haiku alias resolves to, and moves background tasks such as session titles to that model. Never put the key in a project's
.claude/settings.json: that file is committed and shared. -
Start Claude Code and check the route
Run
claudein your project and type/status. You should see the Anthropic base URL line with Uno's address and an Auth token line. Claude Code may print a notice that the model is not in its catalog: Claude Code prints a notice that the model is not in its catalog and keeps the session within 200,000 tokens. Work goes on.$ cd your-project $ claude # in Claude Code: /status
-
Switch per task, or go back to your plan
Run
/model uno/fastfor quick edits and/model uno/smartfor work that needs a plan. To go back to your Claude subscription, remove theenvblock and start Claude Code again: while a gateway key is set, requests carry it instead of your claude.ai login, and your plan's limits do not apply to them.
Check that it works
Give Claude Code a small task that reads and edits a file, then open the Uno console. The minutes appear on the Uno AI page; nothing is taken from your Claude plan.
$ claude -p "Add a comment to the top of README.md that says: checked on Uno AI" --allowedTools "Read,Edit" $ git diff README.md
What each model name costs
| Model you set | What answers | How it is paid |
|---|---|---|
uno/smart | Smart. Today: MiMo v2.6 Pro | Uno AI hours |
uno/fast | Fast. Today: GLM-5.3 Flash | Uno AI hours |
sonnet, opus or another Claude name | The real Claude model | Per token: a Claude model name sent through Uno reaches a real Claude model, billed per token from your premium credit, then your balance |
The short way: Uno Work sets it up for you
On an Uno computer with Uno Work you skip steps 1–4. In Uno Work, Claude Code with no Claude sign-in of its own runs through Uno AI on Claude models, paid from premium credit, then your balance; the free trial asks you to sign in to Claude instead. Sign in with your Claude plan instead, and Claude Code uses that.
Plus: $20 a month
Hours of agent work for Claude Code
Uno AI writes code, builds sites and runs tasks right on your computer. Only the minutes it works count; ten chats in the same minute count as one.
Today: GLM-5.3 Flash (Fast) and MiMo v2.6 Pro (Smart). Dictation into text at no extra cost, within a daily limit.
- 40 hSmall + AIa month
- 100 hPlus + AIa month
- 45 h + FastPro + AIfull speed, then Fast never stops
- UnlimitedMax + AI100 h at full speed, then standard
What it does not do
- It does not make Smart or Fast into Claude. On the Artificial Analysis index Smart has 46, Claude Opus 5.5 has 58.
- Anthropic does not support this setup. A new Claude Code version can change what it sends and break it until Uno catches up.
- Remote Control and voice dictation are off while a gateway key is set.
- On Small+AI and Plus+AI the work stops when the hours end. Only Max+AI is unlimited.
Pick a computer, then add Uno AI
Every plan is a real Linux computer with root. Run Claude Code on it with your Claude plan, with Uno AI, or with both.
Small
A Telegram bot, a VPN, a small backend
$5/ month
- 1 vCPU · 1 GB
- 15 GB disk
- 25 GB cloud storage
- 5 boosts a month
Plus
Agents that work with your laptop closed
$20/ month
- 4 vCPU · 4 GB
- 40 GB disk
- 100 GB cloud storage
- Uno Work in the cloud
Pro
A product in production, a second MacBook
$70/ month
- 16 vCPU · 16 GB
- 120 GB disk
- 300 GB cloud storage
- Speed of a MacBook Air M4
Max
A team and many agents at once
$200/ month
- 32 vCPU · 48 GB
- 300 GB disk
- 1,024 GB cloud storage
- 5 public IPv4
Add Uno AI to any plan: 40 h of AI work a month on Small, 100 h on Plus, unlimited on Max. Today: GLM-5.3 Flash and MiMo v2.6 Pro. Or sign in with your own Claude or ChatGPT.
Common problems
401 "Missing or invalid API key"
ANTHROPIC_AUTH_TOKEN. Run the test request from step 2: it answers the same way.404 "page not found"
/v1. Set ANTHROPIC_BASE_URL to https://console.uno.place with nothing after it./status shows my claude.ai login, not Uno
env block of ~/.claude/settings.json, or run /logout to drop the saved login.The work is billed per token, not in hours
ANTHROPIC_MODEL to uno/smart, and check with /model.