api-key-manager 2.1.0__tar.gz → 2.2.0__tar.gz

{api_key_manager-2.1.0 → api_key_manager-2.2.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: api-key-manager
-Version: 2.1.0
+Version: 2.2.0
 Summary: Batch manage API keys for 44+ AI providers with CLI and Web interfaces
 Author: Townrain
 License: MIT
@@ -51,6 +51,11 @@ Requires-Dist: ruff>=0.4.0; extra == "dev"
 - **SDK 支持** - Python 和 TypeScript 客户端库
 - **Webhook 通知** - 事件驱动的 Webhook 通知系统
 
+## 系统架构
+
+![系统架构流程图](docs/images/flowchart.png)
+
+
 ## 支持的 AI 服务商
 
 ### 国际
@@ -192,10 +197,11 @@ python web.py
 ### 安全防护
 
 - **路径遍历防护** - 导入端点验证路径在允许目录内
-- **SSRF 防护** - `custom_base_url` 验证域名白名单，阻止私有 IP
+- **SSRF 防护** - `custom_base_url` 验证域名白名单，阻止私有 IP，已接入 `check/single` 和 `balance` 端点
 - **时序安全认证** - 使用 `hmac.compare_digest()` 防止时序攻击
 - **认证警告** - 未配置 API Key 时启动警告
 - **密钥掩码** - API 响应中只返回 `key_masked`，不暴露完整密钥
+- **Webhook 安全** - Webhook 端点使用正确的 API 方法，防止运行时错误
 
 ### API 认证
 
@@ -609,7 +615,7 @@ python -m pytest tests/ --cov=key_manager --cov-report=term-missing
 | 密钥解析 | `test_parser.py` | 12 |
 | 验证器 | `test_validator.py` | 5 |
 | 检查器 | `test_checker.py` | 4 |
-| 提供商合约 | `test_providers.py` | 30 |
+| 提供商合约 | `test_providers.py` | 220 |
 | 安全回归 | `test_security.py` | 12 |
 | 加密存储 | `test_storage.py` | 26 |
 | 错误系统 | `test_errors.py` | 28 |
@@ -617,7 +623,12 @@ python -m pytest tests/ --cov=key_manager --cov-report=term-missing
 | 端到端 | `test_e2e.py` | 17 |
 | Webhook | `test_webhook.py` | 35 |
 | OpenAPI | `test_openapi.py` | 26 |
+| 代理检测 | `test_proxy.py` | 19 |
+| 日志系统 | `test_logger.py` | 21 |
+| 能力测试 | `test_tester.py` | 11 |
+| 核心门面 | `test_core.py` | 22 |
 
+**总测试数**: 583+ | **覆盖率**: 88%
 ## SDK 使用
 
 ### Python SDK
@@ -704,6 +715,30 @@ const result = await client.checkSingleKey({ key: 'sk-xxx', provider: 'openai' }
 
 模型列表从 Cherry Studio 同步，每日更新一次。新发布的模型可能需要等待同步后才能被检测到。
 
+
+## 更新日志
+
+### v2.2.0 (2026-06-14)
+
+- **检测逻辑重构**: 添加三步检测逻辑（/v1/models → 并发测试）
+- **URL 修复**: 从 check_endpoint 提取版本路径，修复 OpenCode 等服务商 404 问题
+- **并发优化**: /v1/models 和 chat/completions 都并发调用
+- **超时控制**: 所有网络请求 5 秒超时
+- **模型同步**: 从 Cherry Studio 同步模型数据，支持 ownedBy 映射
+- **新增服务商**: OpenCode Go、OpenCode Zen
+- **新增服务商**: OpenCode Go、OpenCode Zen
+
+### v2.1.2 (2026-06-11)
+
+- **Bug 修复**: 修复 `KeyManager.detect_provider()` 调用异步函数未 await 的问题
+- **测试扩展**: 提供商合约测试扩展至全部 44 个服务商
+- **测试新增**: 新增 `test_proxy.py`、`test_logger.py`、`test_tester.py`、`test_core.py`
+- **覆盖率提升**: 测试覆盖率从 74% 提升至 88%
+
+### v2.1.1
+
+- 初始发布版本
+
 ## 许可证
 
 MIT License

{api_key_manager-2.1.0 → api_key_manager-2.2.0}/README.md

@@ -18,6 +18,11 @@
 - **SDK 支持** - Python 和 TypeScript 客户端库
 - **Webhook 通知** - 事件驱动的 Webhook 通知系统
 
+## 系统架构
+
+![系统架构流程图](docs/images/flowchart.png)
+
+
 ## 支持的 AI 服务商
 
 ### 国际
@@ -159,10 +164,11 @@ python web.py
 ### 安全防护
 
 - **路径遍历防护** - 导入端点验证路径在允许目录内
-- **SSRF 防护** - `custom_base_url` 验证域名白名单，阻止私有 IP
+- **SSRF 防护** - `custom_base_url` 验证域名白名单，阻止私有 IP，已接入 `check/single` 和 `balance` 端点
 - **时序安全认证** - 使用 `hmac.compare_digest()` 防止时序攻击
 - **认证警告** - 未配置 API Key 时启动警告
 - **密钥掩码** - API 响应中只返回 `key_masked`，不暴露完整密钥
+- **Webhook 安全** - Webhook 端点使用正确的 API 方法，防止运行时错误
 
 ### API 认证
 
@@ -576,7 +582,7 @@ python -m pytest tests/ --cov=key_manager --cov-report=term-missing
 | 密钥解析 | `test_parser.py` | 12 |
 | 验证器 | `test_validator.py` | 5 |
 | 检查器 | `test_checker.py` | 4 |
-| 提供商合约 | `test_providers.py` | 30 |
+| 提供商合约 | `test_providers.py` | 220 |
 | 安全回归 | `test_security.py` | 12 |
 | 加密存储 | `test_storage.py` | 26 |
 | 错误系统 | `test_errors.py` | 28 |
@@ -584,7 +590,12 @@ python -m pytest tests/ --cov=key_manager --cov-report=term-missing
 | 端到端 | `test_e2e.py` | 17 |
 | Webhook | `test_webhook.py` | 35 |
 | OpenAPI | `test_openapi.py` | 26 |
+| 代理检测 | `test_proxy.py` | 19 |
+| 日志系统 | `test_logger.py` | 21 |
+| 能力测试 | `test_tester.py` | 11 |
+| 核心门面 | `test_core.py` | 22 |
 
+**总测试数**: 583+ | **覆盖率**: 88%
 ## SDK 使用
 
 ### Python SDK
@@ -671,6 +682,30 @@ const result = await client.checkSingleKey({ key: 'sk-xxx', provider: 'openai' }
 
 模型列表从 Cherry Studio 同步，每日更新一次。新发布的模型可能需要等待同步后才能被检测到。
 
+
+## 更新日志
+
+### v2.2.0 (2026-06-14)
+
+- **检测逻辑重构**: 添加三步检测逻辑（/v1/models → 并发测试）
+- **URL 修复**: 从 check_endpoint 提取版本路径，修复 OpenCode 等服务商 404 问题
+- **并发优化**: /v1/models 和 chat/completions 都并发调用
+- **超时控制**: 所有网络请求 5 秒超时
+- **模型同步**: 从 Cherry Studio 同步模型数据，支持 ownedBy 映射
+- **新增服务商**: OpenCode Go、OpenCode Zen
+- **新增服务商**: OpenCode Go、OpenCode Zen
+
+### v2.1.2 (2026-06-11)
+
+- **Bug 修复**: 修复 `KeyManager.detect_provider()` 调用异步函数未 await 的问题
+- **测试扩展**: 提供商合约测试扩展至全部 44 个服务商
+- **测试新增**: 新增 `test_proxy.py`、`test_logger.py`、`test_tester.py`、`test_core.py`
+- **覆盖率提升**: 测试覆盖率从 74% 提升至 88%
+
+### v2.1.1
+
+- 初始发布版本
+
 ## 许可证
 
 MIT License

{api_key_manager-2.1.0 → api_key_manager-2.2.0}/api_key_manager.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: api-key-manager
-Version: 2.1.0
+Version: 2.2.0
 Summary: Batch manage API keys for 44+ AI providers with CLI and Web interfaces
 Author: Townrain
 License: MIT
@@ -51,6 +51,11 @@ Requires-Dist: ruff>=0.4.0; extra == "dev"
 - **SDK 支持** - Python 和 TypeScript 客户端库
 - **Webhook 通知** - 事件驱动的 Webhook 通知系统
 
+## 系统架构
+
+![系统架构流程图](docs/images/flowchart.png)
+
+
 ## 支持的 AI 服务商
 
 ### 国际
@@ -192,10 +197,11 @@ python web.py
 ### 安全防护
 
 - **路径遍历防护** - 导入端点验证路径在允许目录内
-- **SSRF 防护** - `custom_base_url` 验证域名白名单，阻止私有 IP
+- **SSRF 防护** - `custom_base_url` 验证域名白名单，阻止私有 IP，已接入 `check/single` 和 `balance` 端点
 - **时序安全认证** - 使用 `hmac.compare_digest()` 防止时序攻击
 - **认证警告** - 未配置 API Key 时启动警告
 - **密钥掩码** - API 响应中只返回 `key_masked`，不暴露完整密钥
+- **Webhook 安全** - Webhook 端点使用正确的 API 方法，防止运行时错误
 
 ### API 认证
 
@@ -609,7 +615,7 @@ python -m pytest tests/ --cov=key_manager --cov-report=term-missing
 | 密钥解析 | `test_parser.py` | 12 |
 | 验证器 | `test_validator.py` | 5 |
 | 检查器 | `test_checker.py` | 4 |
-| 提供商合约 | `test_providers.py` | 30 |
+| 提供商合约 | `test_providers.py` | 220 |
 | 安全回归 | `test_security.py` | 12 |
 | 加密存储 | `test_storage.py` | 26 |
 | 错误系统 | `test_errors.py` | 28 |
@@ -617,7 +623,12 @@ python -m pytest tests/ --cov=key_manager --cov-report=term-missing
 | 端到端 | `test_e2e.py` | 17 |
 | Webhook | `test_webhook.py` | 35 |
 | OpenAPI | `test_openapi.py` | 26 |
+| 代理检测 | `test_proxy.py` | 19 |
+| 日志系统 | `test_logger.py` | 21 |
+| 能力测试 | `test_tester.py` | 11 |
+| 核心门面 | `test_core.py` | 22 |
 
+**总测试数**: 583+ | **覆盖率**: 88%
 ## SDK 使用
 
 ### Python SDK
@@ -704,6 +715,30 @@ const result = await client.checkSingleKey({ key: 'sk-xxx', provider: 'openai' }
 
 模型列表从 Cherry Studio 同步，每日更新一次。新发布的模型可能需要等待同步后才能被检测到。
 
+
+## 更新日志
+
+### v2.2.0 (2026-06-14)
+
+- **检测逻辑重构**: 添加三步检测逻辑（/v1/models → 并发测试）
+- **URL 修复**: 从 check_endpoint 提取版本路径，修复 OpenCode 等服务商 404 问题
+- **并发优化**: /v1/models 和 chat/completions 都并发调用
+- **超时控制**: 所有网络请求 5 秒超时
+- **模型同步**: 从 Cherry Studio 同步模型数据，支持 ownedBy 映射
+- **新增服务商**: OpenCode Go、OpenCode Zen
+- **新增服务商**: OpenCode Go、OpenCode Zen
+
+### v2.1.2 (2026-06-11)
+
+- **Bug 修复**: 修复 `KeyManager.detect_provider()` 调用异步函数未 await 的问题
+- **测试扩展**: 提供商合约测试扩展至全部 44 个服务商
+- **测试新增**: 新增 `test_proxy.py`、`test_logger.py`、`test_tester.py`、`test_core.py`
+- **覆盖率提升**: 测试覆盖率从 74% 提升至 88%
+
+### v2.1.1
+
+- 初始发布版本
+
 ## 许可证
 
 MIT License

{api_key_manager-2.1.0 → api_key_manager-2.2.0}/api_key_manager.egg-info/SOURCES.txt

@@ -20,6 +20,7 @@ key_manager/logger.py
 key_manager/model_capabilities.py
 key_manager/parser.py
 key_manager/proxy.py
+key_manager/py.typed
 key_manager/ssrf.py
 key_manager/storage.py
 key_manager/tester.py
@@ -61,6 +62,8 @@ key_manager/providers/modelscope.py
 key_manager/providers/nvidia.py
 key_manager/providers/ocoolai.py
 key_manager/providers/openai.py
+key_manager/providers/opencode.py
+key_manager/providers/opencode_zen.py
 key_manager/providers/openrouter.py
 key_manager/providers/perplexity.py
 key_manager/providers/poe.py
@@ -74,16 +77,24 @@ key_manager/providers/yi.py
 key_manager/providers/zai.py
 key_manager/providers/zhipu.py
 key_manager/providers/zhipu_coding.py
+tests/test_api_endpoints.py
+tests/test_base_check.py
+tests/test_bug_fixes.py
 tests/test_checker.py
+tests/test_core.py
 tests/test_detector.py
 tests/test_e2e.py
 tests/test_errors.py
 tests/test_i18n.py
+tests/test_logger.py
 tests/test_openapi.py
 tests/test_parser.py
 tests/test_provider_detection.py
 tests/test_providers.py
+tests/test_proxy.py
 tests/test_security.py
 tests/test_storage.py
+tests/test_tester.py
 tests/test_validator.py
+tests/test_web_fixes.py
 tests/test_webhook.py

{api_key_manager-2.1.0 → api_key_manager-2.2.0}/key_manager/__init__.py

@@ -1,4 +1,4 @@
-"""Key Manager - Batch API key management for 44+ AI providers."""
+"""Key Manager - Batch API key management for 44+ AI providers."""
 from key_manager.core import KeyManager
 from key_manager.providers import PROVIDERS, get_display_name
 from key_manager.errors import KeyManagerError, ErrorCode
@@ -13,4 +13,5 @@ __all__ = [
     "KeyStore",
 ]
 
-__version__ = "2.1.0"
+__version__ = "2.2.0"
+

{api_key_manager-2.1.0 → api_key_manager-2.2.0}/key_manager/api_models.py

@@ -42,6 +42,7 @@ __all__ = [
     "OperationEntry",
     "OperationsResponse",
     "ProgressResponse",
+    "ProxyResponse",
 ]
 
 
@@ -356,3 +357,11 @@ class ProgressResponse(BaseModel):
     total: int = Field(0, description="Total items to process")
     status: str = Field("", description="Task status: loading / done / error")
     results: dict[str, Any] | None = Field(None, description="Final results when complete")
+
+
+# ── Proxy ─────────────────────────────────────────────────────────────────────
+
+
+class ProxyResponse(BaseModel):
+    proxy: str | None = Field(None, description="Detected proxy URL")
+    source: str = Field("", description="Proxy source: config / env / auto / none")

{api_key_manager-2.1.0 → api_key_manager-2.2.0}/key_manager/core.py

@@ -62,10 +62,16 @@ class KeyManager:
         """Save keys to encrypted storage."""
         self._store.save(data)
     
-    def detect_provider(self, key: str) -> list[str]:
+    async def detect_provider(self, key: str) -> list[str]:
         """Auto-detect provider from key prefix."""
-        return detect_provider(key)
-    
+        import httpx
+        from key_manager.detector import detect_provider as _detect_provider
+        
+        timeout = self.config.get("check", {}).get("timeout_seconds", 30)
+        proxy = self.config.get("proxy") or None
+        async with httpx.AsyncClient(timeout=timeout, proxy=proxy) as client:
+            result = await _detect_provider(client, key)
+            return [result] if result else []
     def list_keys(
         self, 
         provider: str | None = None,
@@ -139,7 +145,7 @@ class KeyManager:
         from key_manager.providers import PROVIDERS
         
         if not provider:
-            providers = self.detect_provider(key)
+            providers = await self.detect_provider(key)
             if not providers:
                 return {"valid": False, "error": "Could not detect provider"}
             provider = providers[0]

{api_key_manager-2.1.0 → api_key_manager-2.2.0}/key_manager/detector.py

@@ -223,31 +223,59 @@ async def detect_provider(client, key: str, suspected_provider: str = None) -> s
         for name, valid in format_results:
             if valid:
                 return name
-    # Step 4: Concurrently probe ALL providers with their top 5 models
+    # Step 4: Concurrently probe ALL providers
+    # First, get models from all providers concurrently
+    async def get_provider_models(name, provider):
+        """Get models from /v1/models endpoint."""
+        try:
+            resp = await asyncio.wait_for(
+                client.get(
+                    f"{provider.get_base_url()}{provider.check_endpoint}",
+                    headers=provider.build_headers(key),
+                ),
+                timeout=5.0
+            )
+            if resp.status_code == 200:
+                data = resp.json()
+                if isinstance(data, dict) and "data" in data:
+                    models = [m.get("id", "") for m in data["data"] if m.get("id")]
+                    if models:
+                        return name, models
+        except:
+            pass
+        return name, []
+    
+    # Get models from all providers concurrently
+    model_tasks = [get_provider_models(name, provider) for name, provider in PROVIDERS.items()]
+    model_results = await asyncio.gather(*model_tasks)
+    
     # Build tasks: (provider_name, model) pairs
     tasks = []
-    for name, provider in PROVIDERS.items():
-        models = PROVIDER_MODELS.get(name, [])
+    for name, models in model_results:
         if not models:
-            models = [getattr(provider, 'check_model', 'gpt-3.5-turbo')]
-        
-        # Use first 5 models
-        for model in models[:5]:
+            continue
+        for model in models:
             tasks.append((name, model))
     
     # Concurrently check all (provider, model) pairs
+    import re
+    
     async def try_model(name, model):
         provider = PROVIDERS[name]
         headers = provider.build_headers(key)
         headers["Content-Type"] = "application/json"
+        # Extract version path from check_endpoint
+        version_match = re.match(r'(/v\d+)', provider.check_endpoint or '')
+        version_prefix = version_match.group(1) if version_match else ''
+        chat_url = f"{provider.get_base_url()}{version_prefix}/chat/completions"
         try:
             resp = await asyncio.wait_for(
                 client.post(
-                    f"{provider.get_base_url()}/chat/completions",
+                    chat_url,
                     headers=headers,
                     json={"model": model, "messages": [{"role": "user", "content": "hi"}], "max_tokens": 5}
                 ),
-                timeout=10.0
+                timeout=5.0
             )
             body = resp.text[:500] if resp.text else ""
             if resp.status_code == 200:

{api_key_manager-2.1.0 → api_key_manager-2.2.0}/key_manager/providers/__init__.py

@@ -43,6 +43,8 @@ from .cstcloud import CSTCloudProvider
 from .zhipu_coding import ZhipuCodingProvider
 from .kimi_coding import KimiCodingProvider
 from .infini_coding import InfiniCodingProvider
+from .opencode import OpenCodeGoProvider
+from .opencode_zen import OpenCodeZenProvider
 
 # Provider registry
 PROVIDERS: dict[str, ProviderBase] = {
@@ -90,6 +92,8 @@ PROVIDERS: dict[str, ProviderBase] = {
     "zhipu-coding": ZhipuCodingProvider(),
     "kimi-coding": KimiCodingProvider(),
     "infini-coding": InfiniCodingProvider(),
+    "opencode-go": OpenCodeGoProvider(),
+    "opencode-zen": OpenCodeZenProvider(),
 }
 
 # Key prefix to provider mapping (for auto-detection)
@@ -113,7 +117,7 @@ KEY_PREFIX_MAP: dict[str, list[str]] = {
     "AKID": ["cstcloud"],
     # Generic sk- prefix (must be last - shared by multiple providers)
     # Excluded: ppio, nvidia, modelscope, ai21 - their /models endpoints don't validate keys
-    "sk-": ["openai", "deepseek", "together", "fireworks", "perplexity", "dashscope", "kimi", "siliconflow", "cerebras", "hyperbolic", "mimo", "stepfun", "infini", "zai", "ai302", "dmxapi", "ocoolai", "dashscope-coding", "tencent-hunyuan"],
+    "sk-": ["openai", "deepseek", "together", "fireworks", "perplexity", "dashscope", "kimi", "siliconflow", "cerebras", "hyperbolic", "mimo", "stepfun", "infini", "zai", "ai302", "dmxapi", "ocoolai", "dashscope-coding", "tencent-hunyuan", "opencode-go", "opencode-zen"],
     # MiniMax Token Plan keys
     "sk-cp-": ["minimax-plan", "infini-coding"],
     # Kimi Coding Plan keys
@@ -228,6 +232,8 @@ PROVIDER_ERROR_SIGNATURES: dict[str, list[str]] = {
     "google": ["generativelanguage"],
     # grok: 实际返回 "Incorrect API key provided: sk***45...console.x.ai."
     "groq": ["groq"],
+    "opencode-go": ["opencode.ai", "zen/go", "creditserror", "no payment method"],
+    "opencode-zen": ["opencode.ai", "zen/v1", "creditserror", "no payment method"],
 }
 
 
@@ -273,6 +279,8 @@ PROVIDER_WEBSITES: dict[str, dict[str, str]] = {
     "longcat": {"name": "LongCat", "url": "https://longcat.com", "docs": "https://longcat.com/docs"},
     "tencent-hunyuan": {"name": "腾讯混元", "url": "https://cloud.tencent.com/product/hunyuan", "docs": "https://cloud.tencent.com/document/product/1729"},
     "cstcloud": {"name": "中算云", "url": "https://www.cstcloud.com", "docs": "https://www.cstcloud.com/docs"},
+    "opencode-go": {"name": "OpenCode Go", "url": "https://opencode.ai", "docs": "https://opencode.ai/docs/zh-cn/go/"},
+    "opencode-zen": {"name": "OpenCode Zen", "url": "https://opencode.ai", "docs": "https://opencode.ai/docs/"},
     "dashscope-coding": {"name": "阿里百炼编程", "url": "https://dashscope.aliyun.com", "docs": "https://help.aliyun.com/zh/dashscope/"},
     "mimo-plan": {"name": "MiMo 计划版", "url": "https://mimo.xiaomi.com", "docs": "https://mimo.xiaomi.com/docs"},
     "minimax-plan": {"name": "MiniMax 计划版", "url": "https://platform.minimaxi.com", "docs": "https://platform.minimaxi.com/document"},

{api_key_manager-2.1.0 → api_key_manager-2.2.0}/key_manager/providers/ai302.py

@@ -26,39 +26,6 @@ class AI302Provider(ProviderBase):
         except Exception:
             return []
 
-    async def check(self, client, key: str) -> CheckResult:
-        """Real usage test - try to make a minimal chat completion request."""
-        headers = self.build_headers(key)
-        headers["Content-Type"] = "application/json"
-        start = time.monotonic()
-        try:
-            resp = await client.post(
-                f"{self.get_base_url()}/chat/completions",
-                headers=headers,
-                json={
-                    "model": "gpt-4o-mini",
-                    "messages": [{"role": "user", "content": "hi"}],
-                    "max_tokens": 5
-                }
-            )
-            latency = (time.monotonic() - start) * 1000
-
-            if resp.status_code == 200:
-                return CheckResult(True, 200, latency, None)
-            elif resp.status_code in (401, 403):
-                return CheckResult(False, resp.status_code, latency, "invalid key or forbidden")
-            elif resp.status_code == 429:
-                return CheckResult(False, 429, latency, "rate limited")
-            else:
-                try:
-                    data = resp.json()
-                    error_msg = data.get("error", {}).get("message", f"status {resp.status_code}")
-                except:
-                    error_msg = f"status {resp.status_code}"
-                return CheckResult(False, resp.status_code, latency, error_msg)
-        except Exception as e:
-            return CheckResult(False, None, (time.monotonic() - start) * 1000, str(e))
-
     async def test_token_limit(self, client, key: str, token_steps: list[int]) -> TestResult:
         headers = self.build_headers(key)
         last_success = None

{api_key_manager-2.1.0 → api_key_manager-2.2.0}/key_manager/providers/baichuan.py

@@ -7,7 +7,6 @@ class BaichuanProvider(ProviderBase):
     name = "baichuan"
     base_url = "https://api.baichuan-ai.com/v1"
     check_endpoint = "/models"
-    check_model = "Baichuan4-Turbo"
 
     def build_headers(self, key: str) -> dict:
         return {"Authorization": f"Bearer {key}"}
@@ -24,33 +23,6 @@ class BaichuanProvider(ProviderBase):
         except Exception:
             return []
 
-    async def check(self, client, key: str) -> CheckResult:
-        headers = self.build_headers(key)
-        headers["Content-Type"] = "application/json"
-        start = time.monotonic()
-        try:
-            resp = await client.post(
-                f"{self.get_base_url()}/chat/completions",
-                headers=headers,
-                json={"model": "Baichuan4-Turbo", "messages": [{"role": "user", "content": "hi"}], "max_tokens": 5}
-            )
-            latency = (time.monotonic() - start) * 1000
-            if resp.status_code == 200:
-                return CheckResult(True, 200, latency, None)
-            elif resp.status_code in (401, 403):
-                return CheckResult(False, resp.status_code, latency, "invalid key or forbidden")
-            elif resp.status_code == 429:
-                return CheckResult(False, 429, latency, "rate limited")
-            else:
-                try:
-                    data = resp.json()
-                    error_msg = data.get("error", {}).get("message", f"status {resp.status_code}")
-                except:
-                    error_msg = f"status {resp.status_code}"
-                return CheckResult(False, resp.status_code, latency, error_msg)
-        except Exception as e:
-            return CheckResult(False, None, (time.monotonic() - start) * 1000, str(e))
-
     async def test_token_limit(self, client, key: str, token_steps: list[int]) -> TestResult:
         headers = self.build_headers(key)
         last_success = None