ApiClo 문서
Choose the exact IDE or client: VS Code, Cursor, Trae, Claude Code, Codex, and more. The correct Base URL, model, and step-by-step guide appear immediately below.
IDE를 선택하고 바로 사용할 설정을 받으세요
이미 사용하는 앱부터 선택하세요. ApiClo가 엔드포인트, 권장 모델, 정확한 가이드를 바로 보여줍니다.
무엇을 연결하고 싶나요?
자율 개발은 OpenCode, DeepSeek Harness 또는 터미널 CLI로 시작하세요. 편집기 확장도 아래에서 사용할 수 있습니다.
Connect Codex CLI/IDE or Claude Code with the config for your OS.
Choose Codex or Claude Code OpenCodePrimary quick path for autonomous project work, tools, and images through an explicit model catalog.
Set up OpenCode DeepSeek HarnessFull agent interface with a custom provider, explicit image input support, and model management.
Set up Harness Oh My PiExtensible power-user CLI with providers, tools, LSP/DAP, and explicit image capabilities.
Set up OMP OrcaManages multiple already configured CLI agents, isolated worktrees, and parallel tasks.
Set up Orca 그냥 채팅하세요브라우저 채팅을 사용하고, UI에서 모델을 선택하고, 사용 가능한 동안 여기에서만 무료 모델을 사용하세요.
오픈채팅 헤르메스 에이전트ApiClo 내 메모리, 웹 컨텍스트, 파일 및 장기 실행 작업이 포함된 작업의 경우.
헤르메스 열기 데스크탑 브리지IDE 또는 에이전트에 로컬 라우터, 폴더 액세스 및 컴퓨터 파일이 필요한 경우 브리지를 다운로드하세요.
.exe 다운로드 Cursor사용자 정의 기본 URL을 통해 채팅/계획에 사용됩니다. 전체 코딩 에이전트 워크플로를 위해서는 Hermes, Cline, Kilo, Roo, Claude Code 또는 Codex를 선호하세요.
커서 설정 자신의 앱SDK, 백엔드, 서비스 및 직접 HTTP 요청을 위한 OpenAI 호환 /v1입니다.
오픈 APIURL과 키를 넣을 위치
짧은 호환성 맵. 클라이언트가 OpenAI 호환 URL을 요청하면 /v1을 사용합니다. Anthropic/Claude 클라이언트가 /v1/messages 자체를 추가하는 경우 /v1 없이 루트를 사용하십시오.
| Client | 최적의 경로 | 기본 URL/키 | 모델 | 상태 |
|---|---|---|---|---|
| 웹 채팅일반 사용자 | ApiClo에서 채팅 열기 | API 키가 필요하지 않습니다 | UI에서 선택됨 | 추천 |
| Hermes계좌 대리인 | 에이전트 → 헤르메스 | ApiClo 계정 | UI에서 선택됨 | 추천 |
| OpenCodeversatile coding start | opencode.json |
https://apiclo.com/v1sk-hub-... |
gpt-6-astracapabilities.input: text + image |
추천 |
| DeepSeek Harnessfine-grained agent setup | profiles/web/cordis.patch.yml |
https://apiclo.com/v1APICLO_API_KEY |
gpt-6-astrainput: text + image |
추천 |
| Oh My PiOMP | ~/.omp/agent/models.yml |
https://apiclo.com/v1APICLO_API_KEY |
gpt-6-astrainput: text + image |
CLI |
| Orcaorchestration | above a configured CLI | configured in the child CLI | inherited from the executor | ADE |
| 데스크탑 브리지로컬 라우터 | 윈도우 .exe 다운로드 | http://127.0.0.1:18441/v1클라이언트 키는 임의의 로컬 값일 수 있습니다. |
Fetch 모델에서 | 추천 |
| 클라인 / 루 / 킬로VS 코드 | OpenAI 호환 프로필 | https://apiclo.com/v1sk-hub-... |
claude-sonnet-5 |
추천 |
| 클로드 코드CLI | ANTHROPIC_AUTH_TOKEN |
https://apiclo.comsk-hub-... |
claude-sonnet-5 |
CLI |
| CodexCLI / IDE extension | ~/.codex/config.toml |
https://apiclo.com/v1APICLAUDE_API_KEY |
claude-sonnet-5 |
CLI |
| Cursorchat/plan | OpenAI-compatible | https://apiclo.com/v1sk-hub-... |
apiclo-gpt-5-6-sol |
제한된 |
| API / SDKHTTP | 오픈AI SDK | https://apiclo.com/v1Authorization: Bearer sk-hub-... |
claude-sonnet-5 |
API |
ApiClo 데스크탑 브리지
로컬 라우터, 폴더 접근 또는 Hermes 권한이 필요할 때만 Desktop Bridge를 설치하세요. IDE와 CLI는 각 가이드에 따라 설정합니다.
파일과 Hermes를 위한 선택적 로컬 라우터 및 컴퓨터 권한 브리지입니다. IDE 설정 설치 프로그램은 아닙니다.
연결 값
이 값은 거의 모든 IDE 및 SDK에서 사용됩니다. 각 프로젝트마다 별도의 키를 만듭니다.
https://apiclo.com/v1
Cursor의 경우 Continue, Kilo, Codex, OpenCode, SDK 및 원시 HTTP입니다.
https://apiclo.com
Claude Code, 원시 Anthropic SDK 및 /v1/messages 자체를 추가하는 클라이언트의 경우.
sk-hub-...
전체 키는 API 키 페이지에 한 번 표시됩니다.
claude-sonnet-5
채팅 및 코드에 대한 균형 잡힌 유료 경로.
무료 Opus/Sonnet/Fable/GPT 별칭은 프로모션 용량이 제공되는 동안 웹 채팅용입니다. 무료* 모델을 IDE, 에이전트 또는 외부 API 클라이언트에 넣지 마십시오. 아래 유료 모델 ID를 사용하세요.
-1m 또는 -1m-context 접미사가 붙은 별도 모델 ID는 더 이상 사용되지 않습니다. 아래의 표준 모델 ID를 선택하세요. 긴 세션은 컨텍스트를 유지하며 필요할 때 자동으로 압축됩니다.
빠른 시작
가입부터 첫 번째 IDE 또는 API 응답까지의 최단 경로입니다.
- 1
Choose your OS: Windows, macOS, or Ubuntu/Linux. Installation commands and config paths below vary by system.
- 2
계정을 만들고 잔액을 추가하세요. 하나의 잔액은 웹 채팅, API, Telegram 및 관리 에이전트에 작동합니다.
- 3
API 키를 열고 특정 IDE 또는 프로젝트에 대한 별도의 키를 만듭니다.
- 4
Codex, Claude Code, and other clients use the separate ready guides below.
- 5
Choose OpenCode for a versatile coding start, DeepSeek Harness for fine-grained agent setup, or Oh My Pi for an extensible CLI with tools.
- 6
For parallel agents and worktrees, use Orca on top of an already configured Codex, Claude Code, OpenCode, or OMP. Orca itself does not replace provider setup.
- 7
짧은 프롬프트로 연결을 테스트하십시오. OK로만 응답하십시오. 그런 다음 파일을 변경하지 않고 현재 폴더 읽기를 테스트합니다.
- 8
Separately attach a small PNG/JPEG and ask what it contains. A text-only OK does not prove the client sent the image to the API.
- 9
긴 코딩 작업의 경우 프로젝트당 하나의 키를 유지하고, 불필요한 모델 전환을 피하고, node_modules, dist, 로그 및 덤프를 컨텍스트에서 제외하세요.
ApiClo accepts OpenAI image_url and Anthropic image blocks. Some IDE and harness clients treat an unknown custom model as text-only and remove the attachment before the request, so the capability must also be declared in the client.
- OpenCode v2:
capabilities: { tools: true, input: [text, image], output: [text] }. Kilo:modalities: { input: [text, image], output: [text] }. - DeepSeek Harness / Oh My Pi / OpenClaw:
input: [text, image]on every model. - Crush:
model add ... --supports-images true. - VS Code native:
vision: true. 계속: enable image input/vision in the model entry used by your version. - Claude Code / Codex: a separate model capability flag is normally unnecessary; attach the file using the native button/command and verify the selected model is vision-capable.
- Aider:
/add image.png,/pasteor launch with the image filename. - Cline / Roo / Cursor / Trae / Goose / Open WebUI: select a vision-capable custom model; if the UI offers Supports images, Vision, or Attachments, enable it.
- Orca: inherits image support from the child CLI it launches; configure vision there first.
Do not add image capability to a text-only model: the client will send the attachment, but upstream will reject it. Test in a new session with a small image and an explicit question about a visible object or text.
IDE agents such as Cline, Kilo, Roo, and Continue often send a system prompt, tool schema, diff, terminal output, and file list. This is normal for coding-agent mode, but these requests cost more than plain chat.
- Start regular coding with Sonnet; use Opus for hard reasoning, architecture review, and difficult bugs.
- Ask the agent to read specific files and folders instead of the whole project. Keep node_modules, dist, logs, dumps, huge JSON/CSV files, and binaries out of context.
- In Continue, keep maxTokens around 1800-4096 for normal chat/edit/apply, and raise it only when a long answer is actually needed.
- For Hermes, keep long work in one chat and avoid unnecessary model switching: memory, cache, and task history work better that way.
In Continue, SDKs, curl, and raw OpenAI-compatible API calls, use the exact id from /v1/models, such as claude-sonnet-5 or claude-opus-4-8. Do not add apiclaude/ before the model id when the client asks for model. Some UIs, such as Kilo, may show provider/model inside their own interface; that is not always the string for direct API calls.
VS 코드 플러그인
Cline, Roo 및 Kilo의 경우 기본적으로 /v1 기본 URL과 함께 OpenAI 호환 프로필을 사용합니다. 클라이언트가 명시적으로 /v1/messages를 호출하는 경우에만 Anthropic 메시지를 기본/고급 모드로 사용하십시오.
Cline
VS Code의 자율 코딩 작업에 적합합니다. ApiClo는 두 가지 방법으로 연결할 수 있습니다.
옵션 A: OpenAI 호환
- 1
VS Code Marketplace에서 Cline을 설치하고 설정/API 구성을 엽니다.
- 2
API 공급자로 OpenAI 호환을 선택합니다.
- 3
/v1: https://apiclo.com/v1을 사용하여 전체 기본 URL을 설정합니다.
- 4
API 키: sk-hub 키입니다. 모델 ID: clude-sonnet-5 또는 clude-opus-4-8.
- 5
프로필을 저장하고 먼저 테스트를 보냅니다. 확인만 답장하세요.
API Provider: OpenAI Compatible
Base URL: https://apiclo.com/v1
API Key: sk-hub-...
Model: claude-sonnet-5
옵션 B: 기본 모드를 위한 인류학적 메시지
- 1
Cline 빌드에 Anthropic 공급자가 필요한 경우 사용자 정의 기본 URL 사용을 활성화합니다.
- 2
사용자 정의 기본 URL: https://apiclo.com . 여기에 /v1을 추가하지 마십시오. Cline은 /v1/messages 자체를 호출합니다.
- 3
API 키: sk-hub 키입니다. 모델: clude-sonnet-5 또는 clude-opus-4-8.
- 4
Anthropic 기본 모드가 불안정한 경우 옵션 A OpenAI 호환으로 돌아갑니다.
Provider: Anthropic
Custom Base URL: https://apiclo.com
Model: claude-sonnet-5
저장 후 보내기: 답장만 OK. 두 번째 테스트: 현재 폴더를 보고 README.md가 존재하는지 확인합니다. 아무것도 변경하지 마십시오.
Images: select a vision-capable model and enable Supports images/Vision if your Cline version exposes that toggle. Start a new task after changing it and attach a PNG with a text question.
모델 21 Model ID
복사 Model ID.
Claude
GPT
루코드
Roo Code는 Cline과 거의 동일한 설정을 사용하지만 공급자 프로필 라벨이 다를 수 있습니다.
- 1
VS Code 확장에서 Roo Code를 설치하고 Roo Code 패널을 엽니다.
- 2
API 키 아래 ApiClo에서 API 키를 생성합니다.
- 3
API 구성/공급자/프로필을 엽니다.
- 4
기본적으로 OpenAI 호환 프로필(기본 URL https://apiclo.com/v1 및 ApiClo 모델 ID)을 선택합니다.
- 5
기본/고급 모드의 경우에만 /v1 없이 Anthropic + 사용자 정의 기본 URL https://apiclo.com를 사용하세요.
- 6
프로필을 저장하고 확인을 확인한 다음 파일 작업을 실행합니다.
- 7
두 번째 테스트: Roo에게 현재 폴더를 읽고 파일을 편집하지 않고 README.md가 존재하는지 말하도록 요청합니다.
OpenAI Compatible Base URL: https://apiclo.com/v1
Anthropic Custom Base URL: https://apiclo.com
Model: claude-opus-4-8
Images: enable Vision/Supports images in the model profile when available. If Roo hides the attachment before the request, verify the same model through OpenCode or Harness with an explicit capability.
모델 21 Model ID
복사 Model ID.
Claude
GPT
킬로 코드
Kilo의 경우 Custom Provider를 사용합니다. Cline/Roo와 달리 기본 URL은 일반적으로 /v1로 입력됩니다.
- 1
VS Code 확장에서 Kilo Code를 설치하고 설정/API 구성을 엽니다.
- 2
ApiClo -> API Keys에서 sk-hub API 키를 받으세요.
- 3
사용자 지정 공급자 만들기: 공급자 ID apiclaude, 표시 이름 ApiClo.
- 4
Claude 모델의 경우 기본적으로 공급자 API: OpenAI 호환 및 기본 URL https://apiclo.com/v1을 선택합니다.
- 5
Kilo가 명시적으로 기본 메시지 모드를 필요로 하는 경우에만 공급자 API: Anthropic 메시지를 사용하세요.
- 6
모델 ID: claude-sonnet-5, claude-opus-4-8 또는 claude-fable-5-1.
- 7
공급자를 저장하고 두 가지 테스트를 실행합니다. 답장만 확인합니다. 그런 다음 편집 없이 프로젝트 파일 목록을 읽습니다.
Provider ID: apiclaude
Provider API: OpenAI Compatible
Base URL: https://apiclo.com/v1
Model ID: claude-sonnet-5
Images: set attachment: true and modalities with input [text, image], output [text] on the custom model. Without modalities, an unknown model is treated as text-only.
"attachment": true,
"modalities": {
"input": ["text", "image"],
"output": ["text"]
}
모델 21 Model ID
복사 Model ID.
Claude
GPT
계속
계속은 config.yaml을 통해 구성됩니다. 모델에게 채팅, 편집, 적용, 요약 역할을 부여합니다.
- 1
계속을 설치하고 구성을 엽니다. 계속 -> 설정 -> config.yaml 열기.
- 2
이미 사용자 정의 모델이 있는 경우 이전 config.yaml을 교체하기 전에 저장하세요.
- 3
아래 블록을 붙여넣고 YOUR_API_KEY를 sk-hub 키로 바꾸세요.
- 4
계속을 다시 시작하거나 IDE에서 창 다시 로드를 실행하세요.
- 5
무거운 작업을 시작하기 전에 작은 파일에 대해 채팅, 편집, 적용을 테스트해 보세요.
~/.continue/config.yaml
name: ApiClo
version: 1.0.0
schema: v1
models:
- name: Claude Sonnet 5
provider: openai
model: claude-sonnet-5
apiBase: https://apiclo.com/v1
apiKey: YOUR_API_KEY
capabilities:
- tool_use
roles:
- chat
- edit
- apply
- summarize
defaultCompletionOptions:
temperature: 0
context:
- provider: code
- provider: docs
- provider: diff
- provider: terminal
- provider: problems
- provider: folder
- provider: codebase
This example does not impose hard contextLength or maxTokens values. Continue uses the selected model's capabilities and shows the active configuration in Continue Console → Options.
In a long Agent session, click Compact conversation, the converging-arrows icon under the latest response. Once context reaches 60%, the same action is available from the context indicator beside the input. Continue keeps a technical summary and sends it with only the newer turns. This compacts history rather than truncating the active task. After compaction, inspect the next request in Continue Console: it should contain the summary and only newer turns, not hundreds of old tool_result blocks.
If Continue rejects provider: openai, use its OpenAI-compatible/custom provider with the same apiBase/apiKey/model. For model, use claude-sonnet-5 or claude-opus-4-8 without the apiclaude/ prefix.
Images: enable image input/vision for the selected custom model in your Continue version. If the field is unavailable, update Continue; a text-only request does not verify attachment delivery.
모델 21 Model ID
복사 Model ID.
Claude
GPT
VS 코드 네이티브 BYOK
VS Code 기본 제공 사용자 지정 끝점의 경우 사용자 구성 파일에 모델을 추가합니다.
- 1
VS Code 사용자 설정 폴더를 열고 chatLanguageModels.json이 없으면 만듭니다.
- 2
아래 JSON을 붙여넣고 YOUR_API_KEY를 sk-hub 키로 바꾸세요.
- 3
URL이 /v1/chat/completions로 끝나는지 확인하세요.
- 4
개발자 실행: 창 새로 고침.
- 5
VS Code 채팅에서 Claude Sonnet 5를 선택하고 Reply only OK 테스트를 보냅니다.
%APPDATA%\Code\User\chatLanguageModels.json
[
{
"name": "ApiClo",
"vendor": "customendpoint",
"apiKey": "YOUR_API_KEY",
"apiType": "chat-completions",
"models": [
{
"id": "claude-sonnet-5",
"name": "Claude Sonnet 5",
"url": "https://apiclo.com/v1/chat/completions",
"toolCalling": true,
"vision": true,
"maxInputTokens": 1000000,
"maxOutputTokens": 64000,
"streaming": true
},
{
"id": "claude-opus-4-8",
"name": "Claude Opus 4.8",
"url": "https://apiclo.com/v1/chat/completions",
"toolCalling": true,
"vision": true,
"maxInputTokens": 1000000,
"maxOutputTokens": 64000,
"streaming": true
}
]
}
]
Developer: Reload Window
모델 21 Model ID
복사 Model ID.
Claude
GPT
IDE 및 데스크톱 클라이언트
OpenAI 호환 설정이 있는 클라이언트는 동일한 기본 URL + API 키 쌍을 사용합니다.
Cursor
Cursor가 내장 모델로 바꾸지 않도록 Cursor 전용 ID를 사용하세요.
- 1
Cursor Settings → Models → API Keys를 열고 OpenAI API Key에 ApiClo sk-hub 키를 저장하세요.
- 2
Override OpenAI Base URL을 켜고 https://apiclo.com/v1 을 정확히 입력하세요.
- 3
아래의 정확한 사용자 지정 ID 중 하나를 추가하세요. /v1/models에 표시되며 원본 모델과 동일한 라우팅, 폴백 및 결제를 사용합니다.
- 4
Cursor 내장 GPT-5.6 Sol 또는 Claude가 아니라 정확한 apiclo-* 항목을 선택하세요.
- 5
macOS와 Windows 설정은 같습니다. Cursor Tab 및 기타 특수 기능은 계속 Cursor 자체 모델을 사용할 수 있습니다.
- 6
'OK만 답해'를 보내세요. ApiClo Logs에 요청이 없으면 Cursor를 업데이트하고 사용자 지정 apiclo-* 모델이 선택되었는지 확인하세요.
OpenAI Base URL: https://apiclo.com/v1
OpenAI API Key: sk-hub-...
Model: apiclo-gpt-5-6-sol
Select a vision-capable custom model and enable Supports images / Vision / Attachments if your Cursor version exposes that switch. If the custom model has no attachment button, that is a client limitation; use OpenCode, Harness, OMP, or Crush with an explicit capability flag.
이 ID는 /v1/chat/completions와 /v1/responses를 모두 지원합니다. 오류가 계속되면 새 Cursor Request ID와 정확한 UTC 시간을 지원팀에 보내세요.
모델 21 Model ID
복사 Model ID.
Claude
GPT
Trae
Trae는 설정 -> 모델 -> 모델 추가 -> 사용자 정의 구성을 통해 연결합니다.
- 1
Trae를 설치하고 로그인하세요. 그렇지 않으면 모델 섹션이 숨겨질 수 있습니다.
- 2
설정 -> 모델 -> 모델 추가 -> 사용자 정의 구성을 엽니다.
- 3
Claude의 경우 API 형식 Anthropic Messages, URL https://apiclo.com 또는 Trae가 전체 /v1 엔드포인트를 요청하는 경우 https://apiclo.com/v1을 선택합니다.
- 4
OpenAI Completions의 경우 https://apiclo.com/v1 및 모델 ID claude-sonnet-5를 사용하세요.
- 5
API 키를 붙여넣고 모델을 저장한 후 Reply only OK 테스트를 실행하세요.
- 6
Trae에 빈 답변이 표시되면 최대 출력을 줄이고 정확한 API 형식을 확인하세요.
API Format: OpenAI Completions
Custom Request URL: https://apiclo.com/v1
Model ID: claude-sonnet-5
Images: select a vision-capable model and enable multimodal/vision/attachments in Custom Config if Trae exposes that option. After saving, always run a separate image + text test.
모델 21 Model ID
복사 Model ID.
Claude
GPT
WebUI / LibreChat 열기
ApiClo를 OpenAI 호환 공급자로 연결합니다.
- 1
자체 호스팅 UI에 대한 공급자 관리 설정 또는 환경 구성을 엽니다.
- 2
OpenAI 호환 공급자를 선택하세요.
- 3
OPENAI_API_BASE_URL=https://apiclo.com/v1 및 OPENAI_API_KEY=sk-hub-...를 설정합니다.
- 4
DEFAULT_MODEL=claude-sonnet-5를 추가하거나 UI에서 모델을 선택하세요.
- 5
UI 컨테이너/프로세스를 다시 시작하고 간단한 테스트를 보냅니다.
OPENAI_API_BASE_URL=https://apiclo.com/v1
OPENAI_API_KEY=sk-hub-...
DEFAULT_MODEL=claude-sonnet-5
Images: in Open WebUI or LibreChat enable vision capability on the custom model/connection. Having an API route alone does not make the UI attach a file.
UI가 /v1 자체를 추가하는 경우 루트 https://apiclo.com를 사용하거나 설정에서 추가 /v1을 제거하세요.
모델 21 Model ID
복사 Model ID.
Claude
GPT
CLI 및 터미널 에이전트
CLI 도구의 경우 OpenAI 호환 변수와 Anthropic 호환 변수를 혼합하지 않는 것이 특히 중요합니다.
OpenCode v2
Recommended quick path: the custom provider lives in opencode.json, and vision is declared in model capabilities.
- 1
Install OpenCode v2 from the official guide: macOS/Linux — npm install -g @opencode/cli; Windows — use the standalone CLI binary from OpenCode.
- 2
Create opencode.json in the project root and paste the configuration below.
- 3
Set APICLO_API_KEY in the OpenCode process environment; do not put the key in the project config.
- 4
Declare each model's actual capabilities. OpenCode v2 assumes text+image for an unknown model, but that is a fallback assumption, not an API check.
- 5
Run opencode, verify a short text response, then attach a small PNG/JPEG and ask it to read visible text.
macOS / Linux:
npm install -g @opencode/cli
{
"$schema": "https://opencode.ai/config.json",
"providers": {
"apiclaude": {
"name": "ApiClo",
"env": ["APICLO_API_KEY"],
"package": "@opencode/ai/providers/openai-compatible",
"settings": {
"baseURL": "https://apiclo.com/v1"
},
"models": {
"gpt-6-astra": {
"name": "GPT 6 Astra",
"capabilities": {
"tools": true,
"input": ["text", "image"],
"output": ["text"]
},
"limit": {
"context": 200000,
"output": 128000
}
},
"claude-fable-5-1": {
"name": "Claude Fable 5.1",
"capabilities": {
"tools": true,
"input": ["text", "image"],
"output": ["text"]
},
"limit": {
"context": 1000000,
"output": 128000
}
}
}
}
},
"model": "apiclaude/gpt-6-astra"
}
opencode
This example targets OpenCode v2. Declare actual capabilities for other models; the v1 config is not interchangeable. On Windows, download the standalone binary below. OpenCode v2
모델 21 Model ID
복사 Model ID.
Claude
GPT
DeepSeek Harness
Harness supports ApiClo as a custom OpenAI-compatible provider. Image input is enabled separately on each model.
- 1
Install DeepSeek Harness from its official instructions. In Settings → Models choose Add model provider → Custom model API.
- 2
Provider ID: apiclo; Display name: ApiClo; Base URL: https://apiclo.com/v1; API protocol: openai-completions.
- 3
Use the APICLO_API_KEY environment variable. If Fetch available models does not list an ID, add the model manually; configure its input types separately.
- 4
In Model options enable Text and Image. In the profile this is input: [text, image] on that model.
- 5
The change applies to the next request without a restart. If a problematic image remains in the session history, create a new session.
- 6
Test text first, then PNG/JPEG. If Harness says model does not support images before the request, the capability still is not saved on the selected model.
For the standard dsh web launch, use $DSH_HOME/profiles/web/cordis.patch.yml; replace web with your profile name otherwise. Preserve other entries when editing an existing file.
$DSH_HOME/profiles/web/cordis.patch.yml
- id: llm-pi-ai
config:
providers:
apiclo:
apiKeyEnv: APICLO_API_KEY
api: openai-completions
baseURL: https://apiclo.com/v1
models:
- id: gpt-6-astra
input: [text, image]
- id: claude-fable-5-1
input: [text, image]
# macOS / Linux
export APICLO_API_KEY="sk-hub-..."
dsh web
# Windows PowerShell
$env:APICLO_API_KEY="sk-hub-..."
dsh web
모델 21 Model ID
복사 Model ID.
Claude
GPT
클로드 코드
Claude Code는 일반적으로 /v1이 없는 Anthropic 호환 기본 URL을 기대합니다.
- 1
Node.js 18+를 확인하세요: node --version.
- 2
클로드 코드 설치: npm install -g @anthropic-ai/claude-code@latest.
- 3
설치를 확인하세요: clude --version.
- 4
ApiClo의 경우 ANTHROPIC_AUTH_TOKEN + ANTHROPIC_BASE_URL을 사용합니다. ANTHROPIC_API_KEY는 x-api-key가 필요한 클라이언트에만 해당됩니다.
- 5
Set CLAUDE_CODE_AUTO_COMPACT_WINDOW=180000 so Claude Code summarizes a long session before the 200k boundary. This does not limit response size.
- 6
먼저 직접 /v1/messages 요청으로 게이트웨이를 확인한 다음 동일한 셸 세션에서 clude를 시작하세요.
- 7
Claude Code에서 /status를 실행하고 ApiClo 기본 URL과 활성 토큰 소스가 표시되는지 확인합니다.
npm install -g @anthropic-ai/claude-code@latest
# macOS / Linux
export ANTHROPIC_AUTH_TOKEN="sk-hub-..."
export ANTHROPIC_BASE_URL="https://apiclo.com"
export ANTHROPIC_MODEL="claude-sonnet-5"
export CLAUDE_CODE_AUTO_COMPACT_WINDOW="180000"
claude
# Windows PowerShell
$env:ANTHROPIC_AUTH_TOKEN="sk-hub-..."
$env:ANTHROPIC_BASE_URL="https://apiclo.com"
$env:ANTHROPIC_MODEL="claude-sonnet-5"
$env:CLAUDE_CODE_AUTO_COMPACT_WINDOW="180000"
claude
~/.claude/settings.json
{
"env": {
"ANTHROPIC_BASE_URL": "https://apiclo.com",
"ANTHROPIC_AUTH_TOKEN": "sk-hub-...",
"CLAUDE_CODE_AUTO_COMPACT_WINDOW": "180000"
}
}
curl /v1/messages smoke
# macOS / Linux
curl -X POST "$ANTHROPIC_BASE_URL/v1/messages" \
-H "Authorization: Bearer $ANTHROPIC_AUTH_TOKEN" \
-H "anthropic-version: 2023-06-01" \
-H "content-type: application/json" \
-d "{\"model\":\"$ANTHROPIC_MODEL\",\"max_tokens\":1,\"messages\":[{\"role\":\"user\",\"content\":\".\"}]}"
# Windows PowerShell
Invoke-RestMethod -Method Post -Uri "$($env:ANTHROPIC_BASE_URL)/v1/messages" `
-Headers @{ "Authorization" = "Bearer $env:ANTHROPIC_AUTH_TOKEN"; "anthropic-version" = "2023-06-01" } `
-ContentType "application/json" `
-Body (@{ model = $env:ANTHROPIC_MODEL; max_tokens = 1; messages = @(@{ role = "user"; content = "." }) } | ConvertTo-Json -Depth 4)
연기가 msg_와 함께 JSON을 반환하면 URL과 키가 작동합니다. 401을 반환하면 클라이언트가 잘못된 헤더가 포함된 키를 보낸 것입니다. x-api-key 시나리오에만 ANTHROPIC_API_KEY를 사용하세요.
Images: in Claude Code attach a local file with the native command/button and select a vision-capable model. A separate provider flag is normally unnecessary, but an image + text smoke is required after changing the endpoint.
모델 21 Model ID
복사 Model ID.
Claude
GPT
Codex
Codex CLI and the IDE extension can use ApiClo through a custom provider in config.toml.
- 1
In the Codex IDE extension, open Codex Settings → Open config.toml; the CLI uses the same ~/.codex/config.toml.
- 2
프로젝트 .codex/config.toml은 공급자/인증을 정의해서는 안 되기 때문에 공급자를 사용자 수준 구성에 추가합니다.
- 3
APICLAUDE_API_KEY를 sk-hub 키로 설정하세요. 이 변수 이름은 아래 env_key에 의해 선택됩니다.
- 4
wire_api = response 유지: 현재 Codex에서는 지원되는 공급자 프로토콜입니다.
- 5
테스트 폴더에서 codex를 실행하고 Reply only OK를 확인합니다.
~/.codex/config.toml
model = "claude-sonnet-5"
model_provider = "apiclaude"
[model_providers.apiclaude]
name = "ApiClo"
base_url = "https://apiclo.com/v1"
env_key = "APICLAUDE_API_KEY"
wire_api = "responses"
APICLAUDE_API_KEY env
# macOS / Linux
export APICLAUDE_API_KEY="sk-hub-..."
codex
# Windows PowerShell
$env:APICLAUDE_API_KEY="sk-hub-..."
codex
Codex가 공급자 구성을 거부하면 먼저 Codex를 업데이트하세요. wire_api를 채팅 완료로 전환하지 마십시오. 현재 Codex 문서에서는 응답을 지원되는 공급자 프로토콜로 설명합니다.
Images: in the IDE extension attach the image through the composer; in CLI provide the local path using the native flow. Test an image + text request separately: provider configuration alone does not prove delivery of the attachment.
모델 21 Model ID
복사 Model ID.
Claude
GPT
OpenClaw
OpenClaw는 두 가지 공급자 모드로 models.json을 통해 연결할 수 있습니다.
- 1
플랫폼에 OpenClaw를 설치하고 openclaw --version을 확인하세요.
- 2
OpenClaw 모델 파일을 찾거나 config 폴더에 models.json을 생성하세요.
- 3
Claude 메시지 모드의 경우 baseUrl https://apiclo.com와 함께 공급자 apiclaude-messages를 사용합니다.
- 4
OpenAI 완성 모드의 경우 baseUrl https://apiclo.com/v1과 함께 공급자 apiclaude-openai를 사용합니다.
- 5
Add input: [text, image] to every vision model, otherwise OpenClaw passes only a text media reference or omits the image.
- 6
Replace apiKey with an sk-hub key and run separate text, tool, and image tests in a project.
{
"models": {
"mode": "merge",
"providers": {
"apiclaude-messages": {
"baseUrl": "https://apiclo.com",
"apiKey": "sk-hub-...",
"auth": "token",
"api": "anthropic-messages",
"models": [
{
"id": "claude-sonnet-5",
"name": "claude-sonnet-5",
"input": ["text", "image"],
"contextWindow": 1000000,
"maxTokens": 64000
}
]
},
"apiclaude-openai": {
"baseUrl": "https://apiclo.com/v1",
"apiKey": "sk-hub-...",
"auth": "token",
"api": "openai-completions",
"models": [
{ "id": "claude-opus-4-8", "name": "claude-opus-4-8", "input": ["text", "image"] }
]
}
}
}
}
모델 21 Model ID
복사 Model ID.
Claude
GPT
퀀 코드
Qwen Code가 OpenAI 호환 클라이언트로 설치되면 OPENAI_* 변수로 충분합니다.
- 1
Node.js를 확인하세요: node --version.
- 2
Qwen 코드 설치: npm install -g @qwen-code/qwen-code@latest.
- 3
qwen --version 및 qwen --help를 확인하세요.
- 4
OpenAI 호환 모드의 경우 OPENAI_API_KEY, OPENAI_BASE_URL 및 OPENAI_MODEL을 설정합니다.
- 5
Qwen Code 빌드가 Anthropic 환경을 지원하는 경우 Claude 모델에 ANTHROPIC_AUTH_TOKEN + https://apiclo.com를 사용할 수 있습니다.
- 6
Use images only if the installed Qwen Code version exposes attachment/image input for a custom endpoint; otherwise the client may send text only.
- 7
qwen을 실행하고 OK/읽기 전용 테스트를 수행합니다.
npm install -g @qwen-code/qwen-code@latest
export OPENAI_API_KEY="sk-hub-..."
export OPENAI_BASE_URL="https://apiclo.com/v1"
export OPENAI_MODEL="claude-sonnet-5"
qwen
$env:OPENAI_API_KEY="sk-hub-..."
$env:OPENAI_BASE_URL="https://apiclo.com/v1"
$env:OPENAI_MODEL="claude-sonnet-5"
qwen
모델 21 Model ID
복사 Model ID.
Claude
GPT
Aider
Aider는 OpenAI 호환 변수를 통해 작동합니다.
- 1
Aider 설치: python -m pip install -U aider-chat.
- 2
테스트 편집이 안전한 git 프로젝트에서 터미널을 엽니다.
- 3
OPENAI_API_BASE 및 OPENAI_API_KEY를 설정합니다.
- 4
aider --model openai/claude-sonnet-5를 실행합니다.
- 5
For an image, run /add image.png or /paste, then ask a text question; the selected model must support vision.
- 6
먼저 파일에 대해 설명해달라고 요청하세요. 연결이 확인된 후에만 편집을 허용합니다.
python -m pip install -U aider-chat
OPENAI_API_BASE=https://apiclo.com/v1
OPENAI_API_KEY=sk-hub-...
aider --model openai/claude-sonnet-5
모델 21 Model ID
복사 Model ID.
Claude
GPT
Oh My Pi (OMP)
OMP connects through ~/.omp/agent/models.yml. Vision capability is declared with the model input field.
- 1
Install OMP with curl -fsSL https://omp.sh/install | sh; on Windows PowerShell use irm https://omp.sh/install.ps1 | iex.
- 2
Create ~/.omp/agent/models.yml and add the apiclo custom provider.
- 3
Set apiKey to the APICLO_API_KEY variable name instead of putting the secret in a repository.
- 4
Keep input: [text, image] on vision models: OMP checks this field before sending an attachment.
- 5
Run omp models find apiclo, then omp and test text, tools, and a separate image.
~/.omp/agent/models.yml
providers:
apiclo:
baseUrl: https://apiclo.com/v1
api: openai-completions
apiKey: APICLO_API_KEY
models:
- id: gpt-6-astra
name: GPT 6 Astra
input: [text, image]
contextWindow: 200000
maxTokens: 128000
- id: claude-fable-5-1
name: Claude Fable 5.1
input: [text, image]
contextWindow: 1000000
maxTokens: 128000
export APICLO_API_KEY="sk-hub-..."
omp models find apiclo
모델 21 Model ID
복사 Model ID.
Claude
GPT
Pi
Pi supports custom OpenAI-compatible models through models.json; declare image capability explicitly.
- 1
Install Pi using its official instructions and create ~/.pi/agent/models.json.
- 2
Set APICLO_API_KEY in the environment; do not store the key in a project file.
- 3
In /model, select apiclo/gpt-6-astra or apiclo/claude-fable-5-1.
- 4
Test a short text request, a tool call, and a separate image with a text question. Declaring input: image alone does not prove delivery.
~/.pi/agent/models.json
{
"providers": {
"apiclo": {
"baseUrl": "https://apiclo.com/v1",
"api": "openai-completions",
"apiKey": "$APICLO_API_KEY",
"models": [
{ "id": "gpt-6-astra", "input": ["text", "image"], "contextWindow": 200000 },
{ "id": "claude-fable-5-1", "input": ["text", "image"], "contextWindow": 1000000 }
]
}
}
}
모델 21 Model ID
복사 Model ID.
Claude
GPT
Kimi Code
Kimi Code accepts a custom OpenAI-compatible provider; ApiClo compatibility is still undergoing our test matrix.
- 1
Install Kimi Code and open ~/.kimi-code/config.toml; preserve your existing settings.
- 2
Set APICLO_API_KEY in the environment and add an openai provider with a /v1 base URL.
- 3
Register the required model with image_in and tool_use, then select it in the client.
- 4
Test text, tools, reasoning, and images separately. Do not treat this integration as validated for autonomous tasks until the tests pass.
~/.kimi-code/config.toml
[providers.apiclo]
type = "openai"
base_url = "https://apiclo.com/v1"
api_key_env = "APICLO_API_KEY"
[models."apiclo/gpt-6-astra"]
provider = "apiclo"
model = "gpt-6-astra"
max_context_size = 200000
capabilities = ["image_in", "tool_use"]
모델 21 Model ID
복사 Model ID.
Claude
GPT
Orca
Orca manages multiple CLI agents and isolated git worktrees. It does not replace the child agent's provider configuration.
- 1
First configure one working ApiClo executor: Codex, Claude Code, OpenCode, or OMP, including a separate vision test.
- 2
Install Orca Desktop/ADE from the official release for your OS and open the target git repository.
- 3
Add the already configured CLI as an agent runtime. The ApiClo key stays in the CLI's secure store or environment, not in Orca project files.
- 4
Use a separate worktree per agent for parallel work and cap the number of concurrent workers.
- 5
Send images to the specific child CLI session. Support is determined by that CLI's model capability: OpenCode modalities, Harness/OMP input, or Crush supports-images.
Orca is an orchestrator, not an OpenAI-compatible provider. Do not put the Base URL and API key into Orca instead of configuring Codex/OpenCode/OMP.
Crush
Crush supports OpenAI-compatible providers and a per-model supports-images flag.
- 1
Install current Crush and open the global crushrc at ~/.config/crush/crushrc or %USERPROFILE%\.config\crush\crushrc.
- 2
Add an apiclo provider of type openai-compat and disable automatic addition of unknown models.
- 3
Register each model with model add and set --supports-images true only for vision models.
- 4
Select the large model with model large and start Crush in a test project.
- 5
On Windows, include a text question with the image: some Crush versions reject an image-only prompt.
~/.config/crush/crushrc
provider add apiclo \
--name "ApiClo" \
--type openai-compat \
--base-url "https://apiclo.com/v1" \
--api-key "$APICLO_API_KEY" \
--discover-models false
model add apiclo/gpt-6-astra \
--name "GPT 6 Astra" \
--context-window 200000 \
--default-max-tokens 128000 \
--can-reason true \
--supports-images true
model large apiclo/gpt-6-astra
모델 21 Model ID
복사 Model ID.
Claude
GPT
Goose
Goose supports custom OpenAI-compatible providers in Desktop and CLI. Verify image capability in the installed version with a separate test.
- 1
Open Settings → Models → Configure providers → Add Custom Provider.
- 2
Provider Type: OpenAI Compatible; Display Name: ApiClo; API URL: https://apiclo.com/v1/chat/completions.
- 3
Enable Authentication, save the sk-hub key, and add gpt-6-astra and claude-fable-5-1.
- 4
Enable Streaming Support and verify a short text request.
- 5
The current Goose custom-provider screen has no universal per-model vision flag. If the attachment button is hidden or the file is removed before the request, use OpenCode, Harness, OMP, or Crush where the capability is explicit.
API URL: https://apiclo.com/v1/chat/completions
Models: gpt-6-astra, claude-fable-5-1
모델 21 Model ID
복사 Model ID.
Claude
GPT
제미니 CLI
Gemini CLI는 빌드에서 사용자 정의 OpenAI/Anthropic 엔드포인트를 설정할 수 없는 한 현재로서는 좋은 직접 ApiClo 대상이 아닙니다.
- 1
먼저 Gemini CLI 버전에 사용자 정의 기본 URL 설정이 있는지 확인하세요.
- 2
커스텀 엔드포인트를 사용할 수 없는 경우 Google 시스템 변수를 재정의하지 말고 Cline, Roo, Kilo 또는 Qwen Code를 사용하세요.
- 3
사용자 지정 OpenAI 호환 엔드포인트를 사용할 수 있는 경우 기본 URL https://apiclo.com/v1, sk-hub API 키 및 모델 claude-sonnet-5를 사용하세요.
- 4
Verify images with a separate test: the custom endpoint must receive an image part, not only a local path. If that Gemini CLI version cannot do this, use OpenCode, Harness, OMP, or Crush.
- 5
설정 후에는 읽기 전용 테스트만 실행하세요. 클라이언트가 여전히 Google 엔드포인트를 호출하는 경우 지원되는 CLI로 전환하세요.
Recommended alternative: Qwen Code, Cline, Roo Code, Kilo Code
에이전트 및 로컬 라우터
관리형 에이전트의 경우 외부 API의 무료 모델 대신 ApiClo 인터페이스를 사용하세요.
헤르메스 에이전트
내장된 Hermes는 ApiClo 내에 있습니다. 균형, 모델, 메모리 및 기록은 에이전트에서 관리됩니다.
- 1
에이전트를 열고 헤르메스를 선택하세요.
- 2
유료 목록에서 모델을 선택하세요: claude-sonnet-5, claude-opus-4-8, claude-sonnet-4-6.
- 3
웹페이지의 경우 채팅에서 웹 컨텍스트를 활성화하세요. URL은 ApiClo 리더/브라우저 대체를 통해 읽혀집니다.
- 4
For an image, use the composer attachment and a vision-capable model; after upload ask a text question to distinguish real image input from a filename.
- 5
Hermes가 컴퓨터에서 작동할 수 있도록 하려면 Desktop Bridge를 다운로드하고 페어링 코드를 만든 다음 필요한 파일 또는 폴더 권한만 부여하세요.
- 6
긴 작업의 경우 이유 없이 체인 중간에 모델을 전환하지 마세요. 캐시와 컨텍스트에 도움이 됩니다.
ApiClo 데스크탑/로컬 라우터
로컬 라우터는 클라이언트가 localhost와만 통신하거나 시스템의 키를 중앙 집중화하려는 경우에 유용합니다.
- 1
ApiClo Desktop Bridge를 다운로드하고 IDE/CLI와 동일한 시스템에서 실행하세요.
- 2
업스트림 sk-hub 키를 데스크톱 브리지에 붙여넣습니다.
- 3
모델 가져오기를 클릭하고 모델 목록이 로드되는지 확인하세요.
- 4
라우터 시작을 클릭합니다.
- 5
IDE/CLI에서 로컬 기본 URL http://127.0.0.1:18441/v1 및 비어 있지 않은 클라이언트 API 키를 설정합니다.
- 6
For images, enable the capability in the IDE/CLI itself: Desktop Router transparently forwards multipart/base64 image content but does not create an attachment button or declare vision for the client.
- 7
테스트 답장만 OK; 응답이 없으면 로컬 라우터가 실행 중이고 포트 18441이 사용 가능한지 확인하십시오.
Download Windows: https://apiclo.com/downloads/ApiCloDesktopBridge.exe
Python app: https://apiclo.com/downloads/desktop-bridge-app.pyw
Python script: https://apiclo.com/downloads/desktop-bridge.py
Local Base URL: http://127.0.0.1:18441/v1
Client API Key: any non-empty value
Upstream key in Desktop Bridge: sk-hub-...
모델 가져오기를 먼저 클릭한 다음 라우터 시작을 클릭합니다. IDE에서는 공개 URL이 아닌 로컬 기본 URL을 사용하세요.
모델 21 Model ID
복사 Model ID.
Claude
GPT
원시 API 및 SDK
ApiClo는 공개 /v1을 통해 OpenAI 호환, Anthropic 호환 및 응답 스타일 요청을 허용합니다.
OpenAI-compatible
IDE, SDK 및 대부분의 자체 호스팅 인터페이스에 대한 기본 경로입니다.
curl https://apiclo.com/v1/chat/completions \
-H "Authorization: Bearer YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "claude-sonnet-5",
"messages": [{"role": "user", "content": "Hi"}],
"max_tokens": 512,
"stream": true
}'
Anthropic-compatible
Claude Messages API를 기대하는 클라이언트에 이 기능을 사용하세요.
curl https://apiclo.com/v1/messages \
-H "x-api-key: YOUR_API_KEY" \
-H "anthropic-version: 2023-06-01" \
-H "Content-Type: application/json" \
-d '{
"model": "claude-sonnet-5",
"max_tokens": 512,
"messages": [{"role": "user", "content": "Hi"}]
}'
응답 API
응답 와이어 API가 필요한 클라이언트에 대해 이를 사용하십시오.
curl https://apiclo.com/v1/responses \
-H "Authorization: Bearer YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "claude-sonnet-5",
"input": "Write a short checklist for release QA."
}'
파이썬 SDK
스크립트, 백엔드 작업 및 테스트에 적합합니다.
from openai import OpenAI
client = OpenAI(
api_key="YOUR_API_KEY",
base_url="https://apiclo.com/v1",
)
response = client.chat.completions.create(
model="claude-sonnet-5",
messages=[{"role": "user", "content": "Hi"}],
max_tokens=512,
)
print(response.choices[0].message.content)
자바스크립트/타입스크립트
프런트엔드 지원 API 경로 및 Node.js 서비스에 적합합니다.
import OpenAI from "openai";
const client = new OpenAI({
apiKey: process.env.APICLAUDE_API_KEY,
baseURL: "https://apiclo.com/v1",
});
const response = await client.chat.completions.create({
model: "claude-sonnet-5",
messages: [{ role: "user", content: "Hi" }],
max_tokens: 512,
});
console.log(response.choices[0].message.content);
Verify before real work
First prove that the client is connected correctly. This takes a minute and saves money on broken IDE requests.
Open the model list with the same key used by the client. If it does not open, fix the key or Base URL first.
https://apiclo.com/v1/models
Send a short non-stream request: Reply only OK. This checks the key, model, and base chat route.
Reply only OK
If the client supports streaming, repeat the same short test with stream. This catches SSE issues before a large task.
stream: true
For an IDE agent, ask it to read the current folder and say whether README.md exists, without editing files.
Look at the current folder and say whether README.md exists. Do not change anything.
If the client uses tools, run a small read-only tool task before heavy coding. Success should include a normal answer, not an empty tool result.
Test long context only after basic checks pass. Start with specific files, not the whole repository.
Keys, usage, and logs
One balance can serve many clients, but separate keys by project make debugging and cost control easier.
Create a separate API key for each IDE, CLI, backend, or test environment. Then Usage and Logs show exactly what spent tokens.
Balance and top-up are in Dashboard. Detailed requests, tokens, cost, status, and errors are in Logs.
Coding extensions often send tool schemas, terminal output, diffs, history, file maps, and a large system prompt. This is normal agent behavior. To reduce spend, ask for specific files, exclude node_modules/dist/logs/dumps, and do not raise maxTokens without a reason.
If you stop a stream after the model has started replying, part of the request may already be counted. Do not blindly retry file-changing tool calls; check Logs and file state first.
After a valid registration through a referral link, the invited user receives one free spin. The referrer does not receive a registration spin; referral rewards are 10% from the first level and 3% from the second level under the referral program rules. Quest spins use a separate prize pool. Every spin wins non-withdrawable ApiClo service credit; current odds are always published in the full rules.
모델
IDE 및 API의 경우 유료 모델 ID를 사용하세요. 웹 채팅에 무료 별칭을 남겨주세요.
GPT 6.1 Sol · API / IDE / Hermes / Web
입력, 출력, 프롬프트 캐시 요금은 GPT 6 Sol과 같습니다. 아래 요금표를 확인하세요.
- API / IDE — API 키로 gpt-6.1-sol을 선택하세요. 아래 예시처럼 /v1/chat/completions 또는 /v1/responses로 요청하세요.
- 웹 채팅 — 모델 선택기에서 GPT 6.1 Sol을 선택하고 메시지를 보내세요. 추론 수준 기본값은 high입니다.
- Hermes — 에이전트 모델 설정이나 채팅 모델 선택기에서 GPT 6.1 Sol을 선택하고 작업을 실행하세요. 도구는 동일한 API 경로를 사용합니다.
gpt-6.1-sol을 명시적으로 선택하세요. 기본값은 high이며 low, medium, high, xhigh, max를 지정할 수 있습니다. none과 minimal은 지원하지 않습니다. 텍스트와 이미지 입력을 지원합니다. 기존 GPT 6 Sol은 별도 모델로 유지됩니다.
IDE와 에이전트 도구에는 ApiClo Chat Completions 또는 Responses를 사용하세요. Chat 도구 호출은 제공자의 Responses로 변환됩니다. 결과에는 원래 tool_call_id 또는 call_id를 사용하고 대화 순서를 유지하세요. 출력 한도는 구독 제공자가 관리합니다.
{"model":"gpt-6.1-sol","reasoning_effort":"high","messages":[{"role":"user","content":"Hello"}],"stream":true}
{"model":"gpt-6.1-sol","reasoning":{"effort":"high"},"input":"Hello","stream":true}
Claude Sonnet 5.5 · API / Hermes / Web
claude-sonnet-5-5를 사용하세요. 추론 수준은 low, medium, high(기본), xhigh, max입니다. Messages는 adaptive와 low/medium/high의 between_tools를 지원합니다. 수동 사고 예산, disabled, 강제 도구 호출, 답변 사전 채우기는 지원하지 않습니다. 도구 반복에서는 서명된 사고 블록과 이전 대화를 그대로 보존하고 다른 모델에 전달하지 마세요. 기본 샘플링 설정을 사용하세요. Claude Code 2.1.284 이상이 필요합니다.
claude-sonnet-5-5
| Model ID | Recommended for | Provider context | Provider output | ApiClo output | 도구 | Vision | Notes |
|---|---|---|---|---|---|---|---|
claude-opus-5-5 |
기본 추론 수준 medium, Opus 5와 동일한 가격 | 1M | 128k | 128k max · 64k default | yes | yes | Use the exact model ID |
gpt-6-sol |
기본 추론 수준 high, Sol 5.6과 동일한 가격 | 1.05M | 128k | 128k max · 64k default | yes | client-dependent | Use the exact model ID |
gpt-6-luna |
Sol 가격의 10분의 1 | 1.05M | 128k | 128k max · 64k default | yes | client-dependent | Use the exact model ID |
gpt-6-astra |
high effort by default; ApiClo input up to 200k | 1.05M | 128k | Responses: provider-managed output limit | yes | client-dependent | Use the exact model ID |
claude-fable-5-1 |
latest Fable version | 1M | 128k | 128k max · 64k default | yes | yes | Use the exact model ID |
gpt-5-6-sol |
flagship preview route | 1.05M | 128k | 128k max · 64k default | yes | client-dependent | Use the exact model ID |
claude-opus-5 |
new flagship route | 1M | 128k | 128k max · 64k default | yes | yes | Use the exact model ID |
claude-sonnet-5-5 |
기본 high; 적응형 사고 | 1M | 128k | 128k max · 64k default | yes | yes | Use the exact model ID |
claude-sonnet-5 |
balanced paid route | 1M | 128k | 128k max · 64k default | yes | yes | Use the exact model ID |
gpt-5-6-terra |
balanced preview route | 1.05M | 128k | 128k max · 64k default | yes | client-dependent | Use the exact model ID |
gpt-5-6-luna |
fast preview route | 1.05M | 128k | 128k max · 64k default | yes | client-dependent | Use the exact model ID |
claude-fable-5 |
previous Fable version | 1M | 128k | 128k max · 64k default | yes | yes | Use the exact model ID |
claude-opus-4-8 |
strong coding and reasoning | 1M | 128k | 128k max · 64k default | yes | yes | Use the exact model ID |
claude-opus-4-7 |
strong coding and reasoning | 1M | 128k | 128k max · 64k default | yes | yes | Use the exact model ID |
claude-sonnet-4-6 |
stable default | 1M | 128k | 128k max · 64k default | yes | yes | Use the exact model ID |
gpt-5-5 |
GPT paid route | 1.05M | 128k | 128k max · 64k default | yes | client-dependent | Use the exact model ID |
claude-haiku-4-5 |
fast and cheaper | 200k | 64k | 64k | yes | yes | Use the exact model ID |
claude-opus-4-6 |
legacy strong route | 1M | 128k | 128k max · 64k default | yes | yes | Use the exact model ID |
claude-sonnet-4-5 |
stable fallback | 200k | 64k | 64k | yes | yes | Use the exact model ID |
claude-opus-4-5 |
legacy strong route | 200k | 64k | 64k | yes | yes | Use the exact model ID |
gpt-6.1-sol |
기본 high; Responses를 통한 도구 호출 | 1.05M | 128k | 출력 한도는 제공자가 관리 | yes | client-dependent | Use the exact model ID |
Context and provider output are the model's official limits. ApiClo no longer applies the old 128k/184k input-token cap: requests are bounded by the selected model's window and the general safe request-size limit. The public API uses a safe 64k output-token default; a client can explicitly request up to 128k on models whose provider limit supports it. On fallback, ApiClo automatically clamps the request to the actual target model's maximum.
Model and prompt-cache pricing
All values are USD per 1 million tokens. Cache rates apply only to tokens explicitly reported by the provider as cache writes or cache reads.
To compare API costs with prompt caching, use the same token mix for every model. The example below uses 1 million total tokens: 400,000 regular input, 400,000 cache-read input, and 200,000 output. The no-cache comparison uses 800,000 regular input and 200,000 output. This is an illustrative scenario, not an observed customer average or a ranking of API providers. Cache writes and refreshes cost extra; your actual price depends on cache hits and model usage.
Formula per 1M total tokens: 0.4 × regular input rate + 0.4 × cache-read rate + 0.2 × output rate.
| 모델 | Regular input | Cache read | Cache write | 산출 | Example per 1M total tokens: with / without cache |
|---|---|---|---|---|---|
Claude Opus 5.5claude-opus-5-5 |
$0.40 | $0.04 · 0.10× |
$0.50 · 1.25× · 5 min $0.80 · 2.00× · 1 h |
$0.40 | $0.256 vs $0.400 |
GPT 6 Solgpt-6-sol |
$0.40 | $0.04 · 0.10× | $0.50 · 1.25× | $0.40 | $0.256 vs $0.400 |
GPT 6 Lunagpt-6-luna |
$0.04 | $0.004 · 0.10× | $0.05 · 1.25× | $0.04 | $0.026 vs $0.040 |
GPT 6 Astragpt-6-astra |
$0.60 | $0.12 · 0.20× | $0.75 · 1.25× | $0.60 | $0.408 vs $0.600 |
Claude Fable 5.1claude-fable-5-1 |
$1.20 | $0.16 · 0.13× |
$1.50 · 1.25× · 5 min $2.40 · 2.00× · 1 h |
$1.20 | $0.784 vs $1.200 |
GPT 5.6 Solgpt-5-6-sol |
$0.40 | $0.04 · 0.10× | $0.50 · 1.25× | $0.40 | $0.256 vs $0.400 |
Claude Opus 5claude-opus-5 |
$0.40 | $0.04 · 0.10× |
$0.50 · 1.25× · 5 min $0.80 · 2.00× · 1 h |
$0.40 | $0.256 vs $0.400 |
Claude Sonnet 5.5claude-sonnet-5-5 |
$0.35 | $0.035 · 0.10× |
$0.4375 · 1.25× · 5 min $0.70 · 2.00× · 1 h |
$0.35 | $0.224 vs $0.350 |
클로드 소네트 5claude-sonnet-5 |
$0.38 | $0.038 · 0.10× |
$0.475 · 1.25× · 5 min $0.76 · 2.00× · 1 h |
$0.38 | $0.243 vs $0.380 |
GPT 5.6 Terragpt-5-6-terra |
$0.30 | $0.03 · 0.10× | $0.375 · 1.25× | $0.30 | $0.192 vs $0.300 |
GPT 5.6 Lunagpt-5-6-luna |
$0.19 | $0.019 · 0.10× | $0.238 · 1.25× | $0.19 | $0.122 vs $0.190 |
클로드 우화 5claude-fable-5 |
$0.80 | $0.08 · 0.10× |
$1.00 · 1.25× · 5 min $1.60 · 2.00× · 1 h |
$0.80 | $0.512 vs $0.800 |
클로드 오푸스 4.8claude-opus-4-8 |
$0.40 | $0.04 · 0.10× |
$0.50 · 1.25× · 5 min $0.80 · 2.00× · 1 h |
$0.40 | $0.256 vs $0.400 |
클로드 오푸스 4.7claude-opus-4-7 |
$0.29 | $0.029 · 0.10× |
$0.363 · 1.25× · 5 min $0.58 · 2.00× · 1 h |
$0.29 | $0.186 vs $0.290 |
클로드 소네트 4.6claude-sonnet-4-6 |
$0.32 | $0.032 · 0.10× |
$0.40 · 1.25× · 5 min $0.64 · 2.00× · 1 h |
$0.32 | $0.205 vs $0.320 |
채팅GPT 5.5gpt-5-5 |
$0.40 | $0.04 · 0.10× | — | $0.40 | $0.256 vs $0.400 |
클로드 하이쿠 4.5claude-haiku-4-5 |
$0.32 | $0.032 · 0.10× |
$0.40 · 1.25× · 5 min $0.64 · 2.00× · 1 h |
$0.32 | $0.205 vs $0.320 |
클로드 오푸스 4.6claude-opus-4-6 |
$0.29 | $0.029 · 0.10× |
$0.363 · 1.25× · 5 min $0.58 · 2.00× · 1 h |
$0.29 | $0.186 vs $0.290 |
클로드 소네트 4.5claude-sonnet-4-5 |
$0.32 | $0.032 · 0.10× |
$0.40 · 1.25× · 5 min $0.64 · 2.00× · 1 h |
$0.32 | $0.205 vs $0.320 |
클로드 오푸스 4.5claude-opus-4-5 |
$0.40 | $0.04 · 0.10× |
$0.50 · 1.25× · 5 min $0.80 · 2.00× · 1 h |
$0.40 | $0.256 vs $0.400 |
GPT 6.1 Solgpt-6.1-sol |
$0.40 | $0.04 · 0.10× | $0.50 · 1.25× | $0.40 | $0.256 vs $0.400 |
Cache reads cost 2/15 of the input rate (about 13.33%) for Claude Fable 5.1, 20% for GPT 6 Astra, and 10% for other models in the current catalog. Five-minute cache writes cost 125%, and Claude one-hour writes cost 200%. If the provider reports no cache-read or cache-write tokens, no cache amount is charged for that category. Automatic fallback keeps the tariff of the model selected by the customer.
Claude reports regular input, cache reads, and cache writes as separate, mutually exclusive provider-usage categories. A request can therefore show zero regular input while the prompt appears under cache read or cache write; those input tokens are not missing. A coding client can create cache again when it changes the cached prefix or sends new cache_control blocks itself.
The exact calculation, cache-read/cache-write token counts, and applied rates for each request are available in Logs → Why charged?
Jev · typed decisions API (beta)
Jev evaluates state and returns typed noul, choice, or score answers. It is not a chat model: /chat/completions, /responses, /messages, streaming, tools, and prose generation are unsupported.
Model ID: jev-latest · POST: https://apiclo.com/v1/decisions · compatible alias https://apiclo.com/v1/systemone · GET: https://apiclo.com/v1/models?type=decision
ApiClo price: $0.0525 per 1M input tokens; output tokens appear in usage but cost $0. Venice beta. Limits: state plus the longest question up to 32K tokens; state plus all questions up to 64K tokens. Evaluate decision quality on your own data.
Send Authorization: Bearer with your ApiClo key, Content-Type: application/json, and a unique Idempotency-Key (8–128 printable ASCII characters) for each logical request. Repeating the same key and body returns the saved answer; a timeout is never automatically resent.
curl -X POST https://apiclo.com/v1/decisions \
-H 'Authorization: Bearer YOUR_APICLO_KEY' \
-H 'Content-Type: application/json' \
-H 'Idempotency-Key: ticket-123-v1' \
-d '{"model":"jev-latest","state":"Payout failed","questions":{"urgent":{"type":"noul","instructions":"Is this urgent?"}}}'
Choice example: {"model":"jev-latest","state":{"ticket":"Duplicate charge"},"questions":{"team":{"type":"choice","instructions":"Which team?","criteria":{"billing":"Payments","technical":"Bugs"}}}}
Score example: {"model":"jev-latest","state":"Service unavailable","questions":{"severity":{"type":"score","instructions":"How severe?","criteria":["Low","Medium","High"]}}}
The TypeSafe Python SDK uses TYPESAFE_BASE_URL=https://apiclo.com and TYPESAFE_API_KEY=your ApiClo key. Pass an Idempotency-Key through extra_headers and disable automatic retries with RetryPolicy(max_retries=0); the SDK calls /v1/systemone.
from typesafe_sdk import Noul, RetryPolicy, TypeSafeClient
with TypeSafeClient(retry=RetryPolicy(max_retries=0)) as client:
result = client.system_one(
"Payout failed three times",
{"urgent": Noul(instructions="Is this urgent?")},
extra_headers={"Idempotency-Key": "ticket-123-v1"},
)
Errors: invalid_decision_request — fix the JSON; balance_too_low — top up ApiClo; provider_credit_exhausted — temporary provider credit issue, no ApiClo charge; provider_rate_limited — Venice limit; provider_completion_unknown — outcome unknown, keep the Idempotency-Key; provider_response_unverified — invalid answer or usage, no charge.
claude-opus-5-5
기본 추론 수준 medium, Opus 5와 동일한 가격 · 1M context · 128k provider output · 128k max · 64k default ApiClo output
gpt-6-sol
기본 추론 수준 high, Sol 5.6과 동일한 가격 · 1.05M context · 128k provider output · 128k max · 64k default ApiClo output
gpt-6-luna
Sol 가격의 10분의 1 · 1.05M context · 128k provider output · 128k max · 64k default ApiClo output
gpt-6-astra
high effort by default; ApiClo input up to 200k · 1.05M context · 128k provider output · Responses: provider-managed output limit ApiClo output
claude-fable-5-1
latest Fable version · 1M context · 128k provider output · 128k max · 64k default ApiClo output
gpt-5-6-sol
flagship preview route · 1.05M context · 128k provider output · 128k max · 64k default ApiClo output
claude-opus-5
new flagship route · 1M context · 128k provider output · 128k max · 64k default ApiClo output
claude-sonnet-5-5
기본 high; 적응형 사고 · 1M context · 128k provider output · 128k max · 64k default ApiClo output
claude-sonnet-5
balanced paid route · 1M context · 128k provider output · 128k max · 64k default ApiClo output
gpt-5-6-terra
balanced preview route · 1.05M context · 128k provider output · 128k max · 64k default ApiClo output
gpt-5-6-luna
fast preview route · 1.05M context · 128k provider output · 128k max · 64k default ApiClo output
claude-fable-5
previous Fable version · 1M context · 128k provider output · 128k max · 64k default ApiClo output
claude-opus-4-8
strong coding and reasoning · 1M context · 128k provider output · 128k max · 64k default ApiClo output
claude-opus-4-7
strong coding and reasoning · 1M context · 128k provider output · 128k max · 64k default ApiClo output
claude-sonnet-4-6
stable default · 1M context · 128k provider output · 128k max · 64k default ApiClo output
gpt-5-5
GPT paid route · 1.05M context · 128k provider output · 128k max · 64k default ApiClo output
claude-haiku-4-5
fast and cheaper · 200k context · 64k provider output · 64k ApiClo output
claude-opus-4-6
legacy strong route · 1M context · 128k provider output · 128k max · 64k default ApiClo output
claude-sonnet-4-5
stable fallback · 200k context · 64k provider output · 64k ApiClo output
claude-opus-4-5
legacy strong route · 200k context · 64k provider output · 64k ApiClo output
gpt-6.1-sol
기본 high; Responses를 통한 도구 호출 · 1.05M context · 128k provider output · 출력 한도는 제공자가 관리 ApiClo output
GPT 6 Astra /v1/responses: max_output_tokens, max_completion_tokens and max_tokens are unsupported on the current route and are not forwarded. X-ApiClo-Ignored-Parameters lists these fields when present. They do not guarantee a maximum response length or cost. JSON and SSE preserve the complete response, history, tools and usage/cache. Billing uses actual usage, which may exceed the requested limit. Use balance and key-access settings for budgeting instead of these fields.
The route retains metadata and user locally for request history and billing, but does not forward them to the provider. Client-supplied fields are also listed in X-ApiClo-Ignored-Parameters. Message contents and tools are unchanged. The client cache key partitions the session and is converted into an attested route key, as with GPT-5.6.
클라이언트에 모델을 찾을 수 없음이 표시되면 먼저 키로 /models를 열고 정확한 ID를 확인하세요.
https://apiclo.com/v1/models
Troubleshooting
일반적인 오류는 거의 항상 키, 기본 URL, 모델 ID 또는 프로토콜 불일치로 인해 발생합니다.
전체 sk-hub 키를 공백 없이 붙여넣었는지 확인하세요. 키가 한 번 표시되었다가 분실된 경우 새 키를 만드세요.
모델 ID를 /models와 비교합니다. ID 대신 표시 이름을 사용하지 마세요.
Most often this is the wrong model id for the client. In Continue, SDKs, and raw API calls, use claude-sonnet-5, not apiclaude/claude-sonnet-5.
클라이언트가 두 번째 /v1을 추가했는지 확인하세요. OpenAI 호환에는 /v1이 필요합니다. Anthropic Messages/custom base에는 일반적으로 /v1이 없는 루트가 필요합니다.
node_modules, dist, 로그, 바이너리 파일 및 긴 덤프를 제외합니다. 긴 작업의 경우 핸드오프 요약을 만듭니다.
백오프로 다시 시도하세요. IDE인 경우 세션 선호도를 유지하려면 동일한 키와 모델 ID를 유지하세요.
일부 IDE는 호환 모드에서 시력을 제한합니다. 먼저 일반 텍스트 요청을 테스트한 다음 이미지를 별도로 테스트하세요.