# APICLAN

> APICLAN (apiclan.us) is an OpenAI-compatible AI API gateway. One base URL
> gives access to Claude, GPT, Gemini, Grok and image models, billed per token
> in USD with no subscription and no monthly minimum. Accounts are topped up
> with USDT; 1 USDT gives 2 credits of API balance. Balance does not expire and
> unused balance can be refunded within 7 days.

## What it is

APICLAN is a paid API relay, not a model provider. Requests are forwarded to
the upstream providers and billed per token at a published fraction of each
model's official list price. The discount comes from the way upstream capacity
is purchased, not from serving different or reduced models — the model, context
window and speed are the provider's own.

## Endpoints

- Base URL for OpenAI-compatible clients: `https://apiclan.us/v1`
- Base URL for Claude Code and the Anthropic SDK: `https://apiclan.us`
  (these clients append `/v1/messages` themselves)
- Authentication: `Authorization: Bearer <key>`, keys start with `sk-`

## Key facts

- Billing is per token, in USD, with no subscription, no monthly minimum and
  no per-request token cap.
- Each API key belongs to exactly one group. A key made for text models cannot
  call image models, and vice versa.
- Cached input is billed at a fraction of the input rate, so agent tools that
  resend the same context cost less than raw token counts suggest.
- For the image models the resolution is set by the suffix on the model name,
  not by the `size` parameter.
- Prices on the site are regenerated automatically from real billing data.

## Troubleshooting OpenAI-compatible APIs

Each page covers one concrete failure, its real cause and a command to
confirm it. Written to be useful on any OpenAI-compatible gateway, not
only this one.

- [Unexpected token '<' when calling an OpenAI-compatible API](https://apiclan.us/help/unexpected-token-openai-api)
- [401 invalid API key — when the key looks right but still fails](https://apiclan.us/help/401-invalid-api-key)
- [404 on /v1/chat/completions when the endpoint clearly exists](https://apiclan.us/help/404-chat-completions)
- [524 timeout on long API requests](https://apiclan.us/help/524-timeout-long-request)
- [Model not found — when the model name is almost right](https://apiclan.us/help/model-not-found)
- [429 from an API gateway means two different things](https://apiclan.us/help/429-quota-vs-rate-limit)
- [CORS error calling an AI API from the browser](https://apiclan.us/help/cors-browser-api-key)
- [400 when forcing a tool call on a reasoning model](https://apiclan.us/help/forced-tool-use-400)
- [Requests die at exactly 100 seconds behind Cloudflare](https://apiclan.us/help/cloudflare-100s-timeout-streaming)
- [The model returns an empty string but you are still billed](https://apiclan.us/help/empty-response-reasoning-max-tokens)
- [A streaming response stops in the middle of a sentence](https://apiclan.us/help/stream-cuts-off-midway)
- [A model that worked yesterday now fails for everyone on the same key group](https://apiclan.us/help/model-works-then-stops-group-routing)
- [The bill is higher than the token count seems to justify](https://apiclan.us/help/bill-higher-than-token-count-suggests)
- [Claude Code and Codex need different base URLs on the same gateway](https://apiclan.us/help/claude-code-codex-baseurl-setup)
- [405 Method Not Allowed when calling an AI API](https://apiclan.us/help/405-method-not-allowed-base-url)
- [404 on /responses — the Responses API needs the version prefix too](https://apiclan.us/help/404-responses-endpoint)
- [context_length_exceeded — what actually counts toward the limit](https://apiclan.us/help/context-length-exceeded)
- [529 overloaded_error from Claude — what it is and what to do](https://apiclan.us/help/529-overloaded-anthropic)
- [401 with a key that is definitely correct — check the header format](https://apiclan.us/help/authorization-header-bearer)
- [400 unsupported parameter on a reasoning model](https://apiclan.us/help/temperature-unsupported-reasoning)
- [Token usage missing from a streamed response](https://apiclan.us/help/stream-usage-missing)
- [400 on a tool call — every tool_use needs a matching tool_result](https://apiclan.us/help/tool-result-block-mismatch)
- [JSON mode still returns text that will not parse](https://apiclan.us/help/json-mode-invalid-output)
- [The call works in curl but not from your application](https://apiclan.us/help/works-in-curl-not-in-app)
- [SSL certificate verify failed when calling an AI API](https://apiclan.us/help/ssl-certificate-verify-failed)
- [socket hang up in the middle of a streamed response](https://apiclan.us/help/socket-hang-up-stream-reset)
- [Claude Code: API Error 400 prompt is too long](https://apiclan.us/help/claude-code-prompt-too-long)
- [Image input rejected by a vision model](https://apiclan.us/help/image-input-invalid-format)
- [The system prompt behaves differently across models](https://apiclan.us/help/system-prompt-role-differences)
- [Reading a 429 properly — retry-after and the limit headers](https://apiclan.us/help/429-retry-after-headers)
- [A custom endpoint's models do not appear in the client's list](https://apiclan.us/help/custom-model-not-listed-in-client)
- [Empty response with finish_reason content_filter](https://apiclan.us/help/content-filter-empty-response)
- [Usage log timestamps do not match when you made the call](https://apiclan.us/help/usage-log-timezone-mismatch)
- [Charged twice for what looked like one request](https://apiclan.us/help/double-charged-on-retry)
- [Concurrency limit exceeded for user, please retry later](https://apiclan.us/help/concurrency-limit-exceeded-for-user)

## Setup and concepts

How to point each client at a gateway, and what the moving parts are.

- [Base URL & API key setup — Claude Code, Codex, Cursor and others](https://apiclan.us/guide)
- [chatbox-custom-api-setup](https://apiclan.us/guides/chatbox-custom-api-setup)
- [cherry-studio-custom-api-setup](https://apiclan.us/guides/cherry-studio-custom-api-setup)
- [Claude API pricing in 2026](https://apiclan.us/guides/claude-api-pricing-2026)
- [claude-code-baseurl](https://apiclan.us/guides/claude-code-baseurl)
- [claude-fable-5-pricing](https://apiclan.us/guides/claude-fable-5-pricing)
- [Claude Opus 5.5 pricing, and what actually changed](https://apiclan.us/guides/claude-opus-5-5-pricing)
- [cline-custom-api-setup](https://apiclan.us/guides/cline-custom-api-setup)
- [codex-custom-api-setup](https://apiclan.us/guides/codex-custom-api-setup)
- [Pointing Claude Desktop and Codex at APICLAN](https://apiclan.us/guides/desktop-apps)
- [gemini-4-argon-api-pricing](https://apiclan.us/guides/gemini-4-argon-api-pricing)
- [GPT-6.1 Sol vs Claude Opus 5.5: same answers, very different bills](https://apiclan.us/guides/gpt-6-1-sol-vs-claude-opus-5-5)
- [gpt-6-astra-api-pricing](https://apiclan.us/guides/gpt-6-astra-api-pricing)
- [gpt-6-astra-vs-gpt-6-1-sol-vs-claude-opus-5-5](https://apiclan.us/guides/gpt-6-astra-vs-gpt-6-1-sol-vs-claude-opus-5-5)
- [GPT-6 Sol vs GPT-6 Astra: what the 5x price gap actually buys](https://apiclan.us/guides/gpt-6-sol-vs-astra)
- [Grok 4.7 API: price, context window and switching from 4.6](https://apiclan.us/guides/grok-4-7-api)
- [What an LLM gateway is, and when you need one](https://apiclan.us/guides/llm-gateway)
- [openai-api-pricing-2026](https://apiclan.us/guides/openai-api-pricing-2026)
- [What “OpenAI-compatible” actually means](https://apiclan.us/guides/openai-compatible-api)
- [opencode-custom-api-setup](https://apiclan.us/guides/opencode-custom-api-setup)
- [Moving from OpenRouter](https://apiclan.us/guides/openrouter-alternative)

## Per-model guides

What each model actually costs here, measured from real billing records.

- [Claude Fable 5.1 (claude-fable-5-1) API pricing, cache rate and the four parameters that return 400](https://apiclan.us/guides/models/claude-fable-5-1)
- [Claude Opus 5 in production](https://apiclan.us/guides/models/claude-opus-5)
- [Claude Sonnet 5 in production](https://apiclan.us/guides/models/claude-sonnet-5)
- [GPT-5.6 Sol in production](https://apiclan.us/guides/models/gpt-5.6-sol)
- [GPT-5.6 Terra in production](https://apiclan.us/guides/models/gpt-5.6-terra)
- [GPT-6 Astra in production](https://apiclan.us/guides/models/gpt-6-astra)
- [GPT Image 2.5, measured](https://apiclan.us/guides/models/gpt-image-2.5)

## Model-to-model price comparisons

Cost structure, not capability claims — where the money actually goes
when you pick one model over another.

- [claude-opus-5 vs claude-sonnet-5](https://apiclan.us/compare/claude-opus-5-vs-claude-sonnet-5)
- [gpt-5.6-sol vs claude-opus-5](https://apiclan.us/compare/claude-opus-5-vs-gpt-5.6-sol)
- [gpt-5.6-terra vs claude-sonnet-5](https://apiclan.us/compare/claude-sonnet-5-vs-gpt-5.6-terra)
- [claude-sonnet-5 vs grok-4.6](https://apiclan.us/compare/claude-sonnet-5-vs-grok-4.6)
- [claude-fable-5-1 vs gpt-6-astra](https://apiclan.us/compare/gpt-6-astra-vs-claude-fable-5-1)
- [claude-sonnet-5 vs gpt-6-sol](https://apiclan.us/compare/gpt-6-sol-vs-claude-sonnet-5)
- [APICLAN vs Kunavo](https://apiclan.us/compare/kunavo)
- [APICLAN vs paying the provider directly](https://apiclan.us/compare/official-api-pricing)
- [APICLAN vs OpenRouter](https://apiclan.us/compare/openrouter)

## Other pages

Reference and company information.

- [Pricing — every model, official price next to ours](https://apiclan.us/pricing)
- [Use cases](https://apiclan.us/use-cases)
- [FAQ](https://apiclan.us/faq)
- [Blog](https://apiclan.us/blog)
- [About](https://apiclan.us/about)
- [Terms, refund policy and contact](https://apiclan.us/legal)

## Languages

Pricing pages, the pricing hub and the image-model guide are published in 14
languages. English lives at the bare path; the others take a language prefix:
zh, zh-Hant, ja, ko, de, fr, es, tr, pt-BR, it, id, vi, ms.

Examples: `https://apiclan.us/pt-BR/pricing`, `https://apiclan.us/ja/pricing/gpt-6-astra`,
`https://apiclan.us/de/guides/models/gpt-image-2.5`.

## Per-model pricing pages

Each model has its own page at `https://apiclan.us/pricing/<model-id>`, with the official price,
the price here, a cost calculator and a worked example from real billing data.

## Operator

APICLAN is operated by NEXUS TECHNOLOGY INNOVATION PTE. LTD., a company incorporated in Singapore
(UEN 202326452W, verifiable in ACRA BizFile). Business address: 60 Paya Lebar Road, #06-28, Paya Lebar Square, Singapore 409051.

## Contact

- Email: support@apiclan.us
- Site: https://apiclan.us