One API. Fewer interruptions.

Route Opus 5, Claude, Codex, GPT, GLM, MiniMax, and DeepSeek through one key with streaming and usage controls.

POST /v1/responses

curl https://openlimits.app/v1/responses \
  -H "Authorization: Bearer $OPENLIMITS_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-5.6-sol",
    "input": "Review and fix this repository"
  }'

…

tokens last 24h

…

requests last 24h

…

cache hit rate

…

avg response time

Recent activity

Past 24 hours · hourly

Requests

…

this hour

Loading hourly activity…

Tokens

…

this hour

Loading hourly activity…

API Cost Savings

…

this hour

Loading hourly activity…

The client stays simple.

OpenLimits handles authentication, routing, failover, and accounting behind compatible provider endpoints.

Request

Use the SDK you already have

Point an Anthropic or OpenAI client at OpenLimits and keep the same request shape.

Route

Match policy to capacity

The router checks plan access, model policy, account health, and current provider availability.

Stream

Return provider output

The response streams to the client while usage and routing metadata are recorded for the dashboard.

Every plan gets the full catalog.

Choose the model for the work, then choose whether you want a day, week, or monthly billing window.

Included with Day, Week, Max, and Ultimate

The current catalog is available regardless of billing period.

Claude Opus 4.8 / 4.7Claude Sonnet 5Claude Haiku 4.5GPT-5.6 Sol / Terra / LunaGPT-5.5 / 5.4 / 5.4 MiniOpus 5GLM 5.3 / 5.2 / 5.1 / 5 TurboMiniMax M3DeepSeek V4 Pro / Flash

Choose the billing window

Day and Week include the same models as Max. Ultimate keeps the monthly format and adds early model access and priority support.

See the model catalog

See what every request used.

A dashboard for routing, not vanity charts.

Inspect live request state, model distribution, timing, token usage, API keys, and billing from one account.

View all features

LiveQueued, retrying, completed, and failed request states.

4 token typesInput, output, cache read, and cache write accounting.

Per modelUsage and cost estimates broken down by requested model.

Per keyFilter activity by the API key used for the request.

Set the endpoint once.

Claude Code and compatible clients can read the base URL and key directly from the environment.

Read the complete setup guide

Shell configuration

export ANTHROPIC_BASE_URL=https://openlimits.app
export ANTHROPIC_AUTH_TOKEN=$OPENLIMITS_KEY

# continue with the normal client
claude "finish the migration"

Choose a pass or a monthly plan.

Every plan includes compatible APIs, routing, streaming, and dashboard analytics.

Day Pass

shortest commitment

€1024 hours

The full current model catalog, billed in a 24-hour window.

Start Day Pass

Included

Opus 5; Claude Opus 4.8 and 4.7, Sonnet 5, Haiku 4.5
GPT-5.6 Sol, Terra, and Luna; GPT-5.5, GPT-5.4, GPT-5.4 Mini
GLM 5.3, GLM 5.2, GLM 5.1, GLM 5 Turbo; MiniMax M3; DeepSeek V4 Pro and Flash
Streaming, tools, images, and thinking
Dashboard analytics and live feed
Renews every 24 hours

Week Pass

€457 days

The full current model catalog, billed in a 7-day window.

Start Week Pass

Included

Opus 5; Claude Opus 4.8 and 4.7, Sonnet 5, Haiku 4.5
GPT-5.6 Sol, Terra, and Luna; GPT-5.5, GPT-5.4, GPT-5.4 Mini
GLM 5.3, GLM 5.2, GLM 5.1, GLM 5 Turbo; MiniMax M3; DeepSeek V4 Pro and Flash
Streaming, tools, images, and thinking
Dashboard analytics and live feed
Renews every 7 days

Secure checkout via StripeCompare all plan details

Common questions

What changes in my setup?

Usually one base URL and one API key. Your client keeps its existing Anthropic or OpenAI request format.

Are there request limits?

The global guard allows 30 request starts per second. Work already in flight is not subject to a concurrency cap.

Which models are included?

Every plan includes Opus 5; Claude Opus 4.8 and 4.7, Sonnet 5, Haiku 4.5; GPT-5.6 Sol, Terra, and Luna; GPT-5.5, GPT-5.4, and GPT-5.4 Mini; GLM 5.3, 5.2, 5.1, and 5 Turbo; MiniMax M3; and DeepSeek V4 Pro and Flash.

Can I inspect usage?

Yes. The dashboard records request state, model, timing, and separate input, output, and cache token counts.

Start with one working day.

A Day Pass costs €10 and activates after checkout.

Start Day Pass View pricing