CroweLM Endpoints

Put CroweLM inside your company.

One endpoint, one key, and a catalog of frontier models available in the tools your team already uses. Works with GitHub Copilot, Cursor, your terminal, or your own code. No shared credit pool to run dry in the middle of a build.

Get a key See the models

Drop-inOpenAI compatible

# point any client at your endpoint
export OPENAI_BASE_URL="https://models.crowelogic.com/v1"
export OPENAI_API_KEY="crowelm-live-..."

# then use it like anything else
curl $OPENAI_BASE_URL/chat/completions \
  -H "Authorization: Bearer $OPENAI_API_KEY" \
  -d '{"model":"crowelm-apex",
       "messages":[{"role":"user",
                    "content":"ship it"}]}'

Five lanes, one interface.

Each lane is tuned to a job rather than a vendor. We route the underlying model, so we can improve what sits behind a lane without you changing a line of configuration.

Apex

crowelm-apex

The top lane. Hard reasoning, architecture decisions, and the problems worth spending real compute on.

Frontier

Core

crowelm-core

The everyday workhorse. Fast enough for agent loops, strong enough for production code.

Balanced

Code

crowelm-code

Purpose-built for writing and editing code. The most economical lane for high-volume agentic work.

Throughput

Reason

crowelm-reason

Extended deliberation for analysis, review, and problems where the first answer is usually wrong.

Deliberate

Flash

crowelm-flash

Low latency for classification, extraction, and the high-frequency calls inside a pipeline.

Fast

Three steps, no migration.

01Point your client

Anything that speaks the OpenAI chat completions format works unchanged. Copilot, Cursor, opencode, the OpenAI SDKs, or a plain curl.

02Use your key

Every key is scoped to the lanes you bought, with its own rate limit, daily ceiling, and monthly ceiling. Your usage is visible at any time.

03Pick a lane

Send a model name and go. Streaming and tool calling are supported on every lane, so agent loops work the same as chat.

Built for teams that ran out of credits.

CapacityNot a shared pool

Your limits are yours. Nobody else's usage throttles you, and nothing goes dark mid-sprint because a company-wide allowance ran out.

ControlCeilings you set

Per-key rate limits and spend ceilings, enforced server side. Set a hard monthly number and it cannot be exceeded by accident.

VisibilityUsage on demand

A usage endpoint returns your plan, your caps, and your current consumption. No waiting for an invoice to find out where you stand.

CompatibilityStreaming and tools

Server-sent events and native tool calling on every lane, verified against real agentic clients rather than assumed.

PortabilityStable names

Lane names stay fixed while the models behind them improve. You upgrade by doing nothing.

PrivacyYour traffic

Requests go to your endpoint with your key. We do not train on your traffic and we do not resell it.

Priced per seat, capped by design.

Every plan includes a monthly usage ceiling enforced at the gateway. If you need more, we raise it deliberately rather than surprising you with a bill.

Solo

For one builder

$150 / month

Core and Code lanes
Your own endpoint and key
Rate and spend ceilings included
Usage endpoint

Start

Studio

For a small team

$250 / month

All five lanes, including Apex
Higher monthly ceiling
Per-seat keys with separate limits
Direct line for support