feat(llm): add 88API as a native LLM provider

.env.example

@@ -350,6 +350,13 @@ MINIMAX_API_KEY=
 MINIMAX_MODEL=MiniMax-M2.7
 MINIMAX_BASE_URL=https://api.minimax.io/v1
 
+# 88API (set LLM_PROVIDER=api88)
+# OpenAI-compatible multi-model gateway: one key for GPT, Claude, Gemini,
+# DeepSeek, Grok, MiniMax and more. Get a key at https://88api.ai
+API88_API_KEY=
+API88_MODEL=claude-sonnet-5-5
+API88_BASE_URL=https://88api.ai/v1
+
 # =========================
 # Common background jobs
 # =========================

README.md

@@ -512,7 +512,7 @@ not include credentials, account data, or exploitable details in public issues.
 | Strategies | Strategy API V2 intents, sizing, risk, backtests, and live runtime. |
 | Crypto | Binance, OKX, Bitget, Bybit, Gate, HTX, and adapter extensions. |
 | Traditional brokers | IBKR and Alpaca workflows. |
-| AI providers | OpenRouter, OpenAI-compatible APIs, Google, DeepSeek, Grok, MiniMax, and custom endpoints. |
+| AI providers | OpenRouter, OpenAI-compatible APIs, Google, DeepSeek, Grok, MiniMax, 88API, and custom endpoints. |
 | Automation | Human API, Agent Gateway, MCP server, Celery jobs, schedules, and notifications. |
 
 Start with the [Indicator guide](docs/trading/INDICATOR_DEV_GUIDE.md),

README_CN.md

@@ -456,7 +456,7 @@ docker compose \
 | 策略 | Strategy API V2 交易意图、仓位计算、风险控制、回测和实盘运行时。 |
 | 加密货币 | Binance、OKX、Bitget、Bybit、Gate、HTX 及可扩展适配器。 |
 | 传统券商 | IBKR 和 Alpaca 工作流。 |
-| AI 提供商 | OpenRouter、OpenAI 兼容 API、Google、DeepSeek、Grok、MiniMax 和自定义端点。 |
+| AI 提供商 | OpenRouter、OpenAI 兼容 API、Google、DeepSeek、Grok、MiniMax、88API 和自定义端点。 |
 | 自动化 | Human API、Agent Gateway、MCP Server、Celery 任务、调度和通知。 |
 
 建议从[指标开发指南](docs/trading/INDICATOR_DEV_GUIDE.md)、

backend_api_python/app/config/api_keys.py

@@ -202,6 +202,16 @@ def LITELLM_API_KEY(cls):
         from app.utils.config_loader import load_addon_config
         val = load_addon_config().get('litellm', {}).get('api_key')
         return val if val else ''
+
+    @property
+    def API88_API_KEY(cls):
+        """88API API key"""
+        env_val = os.getenv('API88_API_KEY', '').strip()
+        if env_val:
+            return env_val
+        from app.utils.config_loader import load_addon_config
+        val = load_addon_config().get('api88', {}).get('api_key')
+        return val if val else ''
 
     @property
     def TAVILY_API_KEYS(cls):

backend_api_python/app/routes/settings.py

@@ -99,7 +99,7 @@
     'AI_CODE_GEN_MODEL', 'LLM_PROXY_URL', 'LLM_USE_SYSTEM_PROXY',
     'JEV_BASE_URL', 'JEV_MODEL', 'JEV_TIMEOUT_SECONDS', 'JEV_MIN_CONFIDENCE', 'AI_DECISION_TIMEOUT_SECONDS',
     'SEC_EDGAR_USER_AGENT',
-    'OPENAI_BASE_URL', 'DEEPSEEK_BASE_URL', 'GROK_BASE_URL', 'ATLASCLOUD_BASE_URL', 'MINIMAX_BASE_URL',
+    'OPENAI_BASE_URL', 'DEEPSEEK_BASE_URL', 'GROK_BASE_URL', 'ATLASCLOUD_BASE_URL', 'MINIMAX_BASE_URL', 'API88_BASE_URL',
     # Trading internals
     'ORDER_MODE', 'MAKER_WAIT_SEC',
     'FINNHUB_API_KEY', 'FINNHUB_FREE_ONLY',
@@ -468,6 +468,7 @@
                     {'value': 'custom', 'label': 'Custom API (OpenAI-compatible)'},
                     {'value': 'minimax', 'label': 'MiniMax'},
                     {'value': 'litellm', 'label': 'LiteLLM (100+ providers)'},
+                    {'value': 'api88', 'label': '88API (Multi-model gateway)'},
                 ],
                 'description': 'Select your preferred LLM provider'
             },
@@ -720,6 +721,35 @@
                 'description': 'Optional. Override provider base URL (e.g. Azure endpoint)',
                 'group': 'litellm'
             },
+            # 88API
+            {
+                'key': 'API88_API_KEY',
+                'label': '88API API Key',
+                'type': 'password',
+                'required': False,
+                'link': 'https://88api.ai/console/token',
+                'link_text': 'settings.link.getApiKey',
+                'description': '88API API key. One key calls GPT, Claude, Gemini, DeepSeek, Grok, MiniMax and more through an OpenAI-compatible endpoint.',
+                'group': 'api88'
+            },
+            {
+                'key': 'API88_MODEL',
+                'label': '88API Model',
+                'type': 'text',
+                'default': 'claude-sonnet-5-5',
+                'link': 'https://88api.ai/pricing',
+                'link_text': 'settings.link.viewModels',
+                'description': '88API model ID without provider prefix, e.g. claude-sonnet-5-5, gpt-5.5, gemini-2.5-pro, deepseek-v4-pro. The API key group must include the model.',
+                'group': 'api88'
+            },
+            {
+                'key': 'API88_BASE_URL',
+                'label': '88API Base URL',
+                'type': 'text',
+                'default': 'https://88api.ai/v1',
+                'description': '88API OpenAI-compatible API endpoint. Must include /v1.',
+                'group': 'api88'
+            },
             # Common settings
             {
                 'key': 'OPENROUTER_TEMPERATURE',

backend_api_python/app/services/llm.py

@@ -1,7 +1,7 @@
 """
 LLM service.
 Supports multiple providers: OpenRouter, OpenAI, Google Gemini, DeepSeek, Grok,
-AtlasCloud, Custom (OpenAI-compatible), MiniMax.
+AtlasCloud, Custom (OpenAI-compatible), MiniMax, 88API.
 Kept separate from AnalysisService to avoid circular imports.
 """
 import json
@@ -115,6 +115,7 @@ class LLMProvider(Enum):
     CUSTOM = "custom"
     MINIMAX = "minimax"
     LITELLM = "litellm"
+    API88 = "api88"  # 88API; enum and env names cannot start with a digit
 
 
 # Provider configurations
@@ -164,6 +165,11 @@ class LLMProvider(Enum):
         "default_model": "openai/gpt-5.4",
         "fallback_model": "gpt-4o-mini",
     },
+    LLMProvider.API88: {
+        "base_url": "https://88api.ai/v1",
+        "default_model": "claude-sonnet-5-5",
+        "fallback_model": "claude-haiku-5-5",
+    },
 }
 
 
@@ -175,7 +181,7 @@ def __init__(self, provider: str = None):
         Initialize LLM service.
 
         Args:
-            provider: Override the default provider (openrouter, openai, google, deepseek, grok, atlascloud, custom, minimax)
+            provider: Override the default provider (openrouter, openai, google, deepseek, grok, atlascloud, custom, minimax, api88)
         """
         self._provider_override = provider
         # Populated by call_llm_api() so callers can report what actually answered;
@@ -207,7 +213,7 @@ def provider(self) -> LLMProvider:
                 pass
 
         # Auto-detect: find any provider with a configured API key
-        # Priority: DeepSeek > AtlasCloud > Grok > MiniMax > OpenAI > Google > OpenRouter
+        # Priority: DeepSeek > AtlasCloud > Grok > MiniMax > OpenAI > Google > 88API > OpenRouter
         # (LiteLLM excluded from auto-detect; must be set explicitly via LLM_PROVIDER=litellm)
         priority_order = [
             LLMProvider.DEEPSEEK,
@@ -216,6 +222,7 @@ def provider(self) -> LLMProvider:
             LLMProvider.MINIMAX,
             LLMProvider.OPENAI,
             LLMProvider.GOOGLE,
+            LLMProvider.API88,
             LLMProvider.OPENROUTER,
         ]
 
@@ -241,6 +248,7 @@ def get_api_key(self, provider: LLMProvider = None) -> str:
             LLMProvider.CUSTOM: APIKeys.CUSTOM_API_KEY,
             LLMProvider.MINIMAX: APIKeys.MINIMAX_API_KEY,
             LLMProvider.LITELLM: APIKeys.LITELLM_API_KEY,
+            LLMProvider.API88: APIKeys.API88_API_KEY,
         }
         return key_map.get(p, "") or ""
 
@@ -447,7 +455,7 @@ def _llm_post(self, url: str, *, headers: dict, json_payload: dict, timeout: int
     def _call_openai_compatible(self, messages: list, model: str, temperature: float, 
                                  api_key: str, base_url: str, timeout: int,
                                  use_json_mode: bool = True) -> str:
-        """Call OpenAI-compatible API (OpenAI, DeepSeek, Grok, AtlasCloud, OpenRouter)."""
+        """Call OpenAI-compatible API (OpenAI, DeepSeek, Grok, AtlasCloud, OpenRouter, 88API)."""
         url = f"{base_url}/chat/completions"
 
         headers = {"Content-Type": "application/json"}
@@ -470,18 +478,25 @@ def _call_openai_compatible(self, messages: list, model: str, temperature: float
         # its public parameter table currently lists model/messages/temperature/
         # max_tokens/stream/top_p and not response_format. Keep prompts JSON-
         # oriented while avoiding a provider-side 400 from an unsupported knob.
-        if use_json_mode and "atlascloud" not in (base_url or "").lower():
+        # 88API routes claude-* models to Anthropic, which rejects
+        # {"type": "json_object"} (only text or json_schema), so skip it there too.
+        normalized_base_url = (base_url or "").lower()
+        skip_json_mode = "atlascloud" in normalized_base_url or (
+            "88api.ai" in normalized_base_url and (model or "").lower().startswith("claude")
+        )
+        if use_json_mode and not skip_json_mode:
             data["response_format"] = {"type": "json_object"}
 
         response = self._llm_post(url, headers=headers, json_payload=data, timeout=timeout)
 
         # Handle non-2xx with provider/model-aware details
         if response.status_code >= 400:
-            normalized_base_url = (base_url or "").lower()
             if "atlascloud" in normalized_base_url:
                 provider_name = "AtlasCloud"
             elif "openrouter" in normalized_base_url:
                 provider_name = "OpenRouter"
+            elif "88api.ai" in normalized_base_url:
+                provider_name = "88API"
             else:
                 provider_name = "LLM"
             err_text = self._extract_provider_error(response)
@@ -1137,6 +1152,8 @@ def _normalize_model_for_provider(self, model: str, provider: LLMProvider) -> st
                 'atlascloud': LLMProvider.ATLASCLOUD,
                 'atlas': LLMProvider.ATLASCLOUD,
                 'minimax': LLMProvider.MINIMAX,
+                'api88': LLMProvider.API88,
+                '88api': LLMProvider.API88,
             }
 
             # If the model prefix matches the current provider, use the extracted model name
@@ -1171,6 +1188,8 @@ def _detect_provider_from_model(self, model: str) -> Optional[LLMProvider]:
             'atlascloud': LLMProvider.ATLASCLOUD,
             'atlas': LLMProvider.ATLASCLOUD,
             'minimax': LLMProvider.MINIMAX,
+            'api88': LLMProvider.API88,
+            '88api': LLMProvider.API88,
             'anthropic': LLMProvider.OPENROUTER,  # Anthropic only via OpenRouter
             'meta': LLMProvider.OPENROUTER,  # Meta/Llama only via OpenRouter
             'mistral': LLMProvider.OPENROUTER,  # Mistral only via OpenRouter
@@ -1247,7 +1266,7 @@ def call_llm_api(self, messages: list, model: str = None, temperature: float = 0
                 )
             # If no API key for current provider, try to find any available provider
             if try_alternative_providers:
-                for alt_provider in [LLMProvider.DEEPSEEK, LLMProvider.ATLASCLOUD, LLMProvider.GROK, LLMProvider.MINIMAX, LLMProvider.OPENAI, LLMProvider.GOOGLE, LLMProvider.OPENROUTER]:
+                for alt_provider in [LLMProvider.DEEPSEEK, LLMProvider.ATLASCLOUD, LLMProvider.GROK, LLMProvider.MINIMAX, LLMProvider.OPENAI, LLMProvider.GOOGLE, LLMProvider.API88, LLMProvider.OPENROUTER]:
                     if alt_provider != p and self.get_api_key(alt_provider):
                         logger.warning(f"No API key for {p.value}, switching to {alt_provider.value}")
                         p = alt_provider
@@ -1460,7 +1479,7 @@ def _try_alternative_providers(self, messages: list, model: str, temperature: fl
         """
         Try alternative providers when current provider fails.
 
-        Priority: DeepSeek > AtlasCloud > Grok > MiniMax > OpenAI > Google > OpenRouter
+        Priority: DeepSeek > AtlasCloud > Grok > MiniMax > OpenAI > Google > 88API > OpenRouter
         """
         priority_order = [
             LLMProvider.DEEPSEEK,
@@ -1469,6 +1488,7 @@ def _try_alternative_providers(self, messages: list, model: str, temperature: fl
             LLMProvider.MINIMAX,
             LLMProvider.OPENAI,
             LLMProvider.GOOGLE,
+            LLMProvider.API88,
             LLMProvider.OPENROUTER,
         ]
 

backend_api_python/app/utils/config_loader.py

@@ -138,6 +138,11 @@ def env_get(name: str) -> Optional[str]:
         ('LITELLM_BASE_URL', 'litellm.base_url', 'string'),
         ('LITELLM_MODEL', 'litellm.model', 'string'),
 
+        # 88API
+        ('API88_API_KEY', 'api88.api_key', 'string'),
+        ('API88_BASE_URL', 'api88.base_url', 'string'),
+        ('API88_MODEL', 'api88.model', 'string'),
+
         # LLM Provider Selection
         ('LLM_PROVIDER', 'llm.provider', 'string'),
         ('LLM_PROXY_URL', 'llm.proxy_url', 'string'),

backend_api_python/env.example

@@ -229,6 +229,13 @@ MINIMAX_API_KEY=
 MINIMAX_MODEL=MiniMax-M2.7
 MINIMAX_BASE_URL=https://api.minimax.io/v1
 
+# 88API (set LLM_PROVIDER=api88)
+# OpenAI-compatible multi-model gateway: one key for GPT, Claude, Gemini,
+# DeepSeek, Grok, MiniMax and more. Get a key at https://88api.ai
+API88_API_KEY=
+API88_MODEL=claude-sonnet-5-5
+API88_BASE_URL=https://88api.ai/v1
+
 # =========================
 # Common background jobs
 # =========================