Can I use Claude Code without an Anthropic Max plan?

Yes. Claude Code is the harness. LayerX1 is the backend. Set ANTHROPIC_BASE_URL to https://api.layerx1.com and ANTHROPIC_AUTH_TOKEN to your lx1_ key. You are billed by LayerX1, not Anthropic.

Does GLM work inside Claude Code?

Yes. Point Claude Code at LayerX1 and set ANTHROPIC_MODEL to a GLM id from the catalog (for example lx1-glm-5.2). The Messages dialect is native — no local proxy.

How is this cheaper than Claude Code Max?

Anthropic Max is a seat on Anthropic's own models. A LayerX1 coding plan is a monthly usage pool at public list rates across GLM, MiniMax, Kimi, DeepSeek, Qwen and flagship models, from $5. Heavy agent loops cost less on the open coding models; you still have Claude when a turn needs it.

Is this the same as OpenRouter for Claude Code?

OpenRouter is a pay-as-you-go supermarket of hundreds of models. LayerX1 is a coding plan: a flat monthly pool, an installer that writes Claude Code / Codex / Cline config, and an API built for hour-long agent loops. Same ANTHROPIC_BASE_URL trick, different product.

What do I set for ANTHROPIC_BASE_URL?

https://api.layerx1.com — the origin only. Do not append /v1. Claude Code adds /v1/messages itself. Put the key in ANTHROPIC_AUTH_TOKEN, not ANTHROPIC_API_KEY.

Claude Code

Run Claude Code cheaper.
Same harness.

Claude Code is the agent. LayerX1 is the backend. Four environment variables — or one command — and you keep the loop on GLM, MiniMax, Kimi, DeepSeek or Claude, on a $5 coding plan instead of a $100 Max seat.

Get a key See plans

Free $2 credit · no card · npx layerx1 setup --tool claude-code

Setup[01/04]

Point Claude Code here. Keep the harness.

Claude Code speaks the Anthropic Messages dialect. LayerX1 serves that dialect natively — no local proxy, no Claude Code Router. The base URL is the origin only; Claude Code appends /v1/messages itself.

ONE COMMAND

npx layerx1 setup --tool claude-code

OR FOUR VARIABLES

export ANTHROPIC_BASE_URL=https://api.layerx1.com
export ANTHROPIC_AUTH_TOKEN=lx1_your_key
export ANTHROPIC_MODEL=lx1-glm-5.2
export ANTHROPIC_SMALL_FAST_MODEL=lx1-glm-4.7-flash

01
Get a key
Sign up at layerx1.com. The free plan includes a small credit. Paid coding plans start at $5/month.
02
Point Claude Code at LayerX1
Run npx layerx1 setup --tool claude-code, or set ANTHROPIC_BASE_URL to https://api.layerx1.com and ANTHROPIC_AUTH_TOKEN to your lx1_ key.
03
Pick a model
Set ANTHROPIC_MODEL to a catalog id such as lx1-glm-5.2 or lx1-sonnet-4.6. Set ANTHROPIC_SMALL_FAST_MODEL to a cheap flash model for helper calls.
04
Start Claude Code
Run claude. If you were logged into Anthropic before, run /logout once so the gateway token is used.

Full Claude Code guide →

Against the shelf[02/04]

The buyer is already shopping coding plans.

GLM Lite, MiniMax Token Plan, OpenCode Go, ClinePass, OpenRouter — that is the shelf. LayerX1 sits on it as the one key that is not locked to a single lab or a single harness.

Plan	Entry	What you actually get
Anthropic Max	$100–$200/mo	Claude only. Seat + usage windows. Hits a wall mid-sprint.
GLM / MiniMax coding plan	$10–$18/mo	One lab's models. Fine until you want Claude, GPT, or a second harness.
OpenRouter	PAYG + fee	Every model, no installer, no monthly pool. A supermarket, not a coding plan.
LayerX1	$5 / $19 / $49	One key. Claude Code native. GLM, MiniMax, Kimi, DeepSeek, Qwen, Claude, GPT. Sixteen harnesses.

Coding plans compared →

Models that hold a loop[03/04]

Pick the model. Keep the agent.

Swap ANTHROPIC_MODEL any time. Helper calls stay on a flash model so the expensive id only spends on the turns that need it.

lx1-glm-5.2

Daily driver — GLM in Claude Code, coding-plan price.

lx1-minimax-m3

Long context, high tokens/sec agent loops.

lx1-kimi-k2.7-code

Moonshot coding weights, same harness.

lx1-sonnet-4.6

Anthropic when the turn is actually hard.

Still signed into Anthropic from before? Run /logout once so Claude Code picks up ANTHROPIC_AUTH_TOKEN. 404s usually mean you appended /v1 to the base URL. Codex CLI uses the same key on the OpenAI dialect.

Frequently asked[04/04]

Can I use Claude Code without an Anthropic Max plan?: Yes. Claude Code is the harness. LayerX1 is the backend. Set ANTHROPIC_BASE_URL to https://api.layerx1.com and ANTHROPIC_AUTH_TOKEN to your lx1_ key. You are billed by LayerX1, not Anthropic.
Does GLM work inside Claude Code?: Yes. Point Claude Code at LayerX1 and set ANTHROPIC_MODEL to a GLM id from the catalog (for example lx1-glm-5.2). The Messages dialect is native — no local proxy.
How is this cheaper than Claude Code Max?: Anthropic Max is a seat on Anthropic's own models. A LayerX1 coding plan is a monthly usage pool at public list rates across GLM, MiniMax, Kimi, DeepSeek, Qwen and flagship models, from $5. Heavy agent loops cost less on the open coding models; you still have Claude when a turn needs it.
Is this the same as OpenRouter for Claude Code?: OpenRouter is a pay-as-you-go supermarket of hundreds of models. LayerX1 is a coding plan: a flat monthly pool, an installer that writes Claude Code / Codex / Cline config, and an API built for hour-long agent loops. Same ANTHROPIC_BASE_URL trick, different product.
What do I set for ANTHROPIC_BASE_URL?: https://api.layerx1.com — the origin only. Do not append /v1. Claude Code adds /v1/messages itself. Put the key in ANTHROPIC_AUTH_TOKEN, not ANTHROPIC_API_KEY.

Put your agent on inference built for the work

Your agent stays the same.
Its inference gets better.

$export ANTHROPIC_BASE_URL=https://api.layerx1.com

Get a key Read the docs

Start free · no card · Starter from $5/mo

Point Claude Code here. Keep the harness.

Get a key

Point Claude Code at LayerX1

Pick a model

Start Claude Code

The buyer is already shopping coding plans.

Pick the model. Keep the agent.

Your agent stays the same.Its inference gets better.

Your agent stays the same.
Its inference gets better.