--- title: "推薦設定" url: "/docs/auto_router/recommended_configurations" canonical_url: "https://docs.litellm.ai/docs/auto_router/recommended_configurations" type: "docs" last_updated: "2026-10-08" summary: "將內建的 1M Context、Anthropic、OpenAI、Gemini 與 Lite 預設值整理為 config.yaml、這些設定所對應的公開基準測試配置，以及如何在它們之間選擇。" related: - "/docs/auto_router/user_setup" - "/docs/auto_router/benchmarks" --- # 推薦設定 > Index of all LiteLLM docs: https://docs.litellm.ai/llms.txt 建議的梯級，與儀表板的 Auto Router 範本相符。每個層級都必須在同一個檔案中存在為 `model_name`；將提供者前綴或憑證替換為您自己的部署，router 項目就會保持不變。 | 梯級 | SIMPLE | MEDIUM | COMPLEX | REASONING | 分類器 | | --- | --- | --- | --- | --- | --- | | [1M Context](#1m-context) | gpt-5.6-luna | gpt-5.6-terra | gpt-5.6-sol | claude-opus-5, high effort | heuristic v2 | | [Anthropic Family](#anthropic-family) | claude-haiku-4-5 | claude-sonnet-5 | claude-opus-5 | claude-opus-5, high effort | heuristic | | [OpenAI Family](#openai-family) | gpt-5.6-luna | gpt-5.6-terra | gpt-5.6-sol | gpt-5.6-sol, xhigh effort | heuristic | | [Gemini Family](#gemini-family) | gemini-2.5-flash-lite | gemini-3.1-flash-lite | gemini-3.7-flash | gemini-3.1-pro-preview | heuristic | | [Lite](#lite) | deepseek-v4-flash | muse-spark-1.2, xhigh | kimi-k3, max | claude-opus-5 | LLM, agentic rubric | | [Benchmark config](#the-benchmark-configuration) | claude-haiku-4-5 | claude-sonnet-5 | claude-opus-5 | claude-opus-5 | LLM, gpt-5.4-mini | | [Production config](#the-production-configuration) | claude-haiku-4-5 | claude-haiku-4-5 | claude-sonnet-5 | claude-opus-5 | heuristic | ## 選擇梯級 {#choosing-a-ladder} - **單一家族**：當用戶端依賴提供者特定行為（Anthropic 快取控制、OpenAI reasoning 參數）時。每個層級都維持在同一個 API 表面。 - **Lite**：當成本比提供者一致性更重要且流量是 agentic 時。混合提供者，使用帶有 `agentic` rubric 的 LLM 分類器。 - **1M Context**：適用於長提示。Luna、Terra、Sol，然後在高 effort 時用 Opus 5，搭配 heuristic v2 分類器。 - **相同模型，更多 effort**：適用於最上層。增加的是輸出 token 成本，而不是更高的每 token 費率。模式：[effort 梯級](/docs/proxy/auto_routing#effort-ladders)。 - 所有梯級都會關閉 `session_affinity`（預設；請參閱 [prompt caching](/docs/auto_router/prompt_caching)），並設定 `escalation_keywords: ["LITELLM ESCALATE"]`；當完全相同的片語出現時，會將請求提升一個層級。比對區分大小寫。 ## Anthropic Family {#anthropic-family} Haiku、Sonnet、Opus，然後在高 reasoning effort 時使用 Opus。 ```yaml title="config.yaml" keep-model-ids model_list: - model_name: claude-haiku-4-5 litellm_params: model: anthropic/claude-haiku-4-5 api_key: os.environ/ANTHROPIC_API_KEY - model_name: claude-sonnet-5 litellm_params: model: anthropic/claude-sonnet-5 api_key: os.environ/ANTHROPIC_API_KEY - model_name: claude-opus-5 litellm_params: model: anthropic/claude-opus-5 api_key: os.environ/ANTHROPIC_API_KEY - model_name: claude-opus-5-high litellm_params: model: anthropic/claude-opus-5 api_key: os.environ/ANTHROPIC_API_KEY reasoning_effort: high - model_name: claude-auto litellm_params: model: auto_router/complexity_router complexity_router_config: tiers: SIMPLE: claude-haiku-4-5 MEDIUM: claude-sonnet-5 COMPLEX: claude-opus-5 REASONING: claude-opus-5-high classifier_type: heuristic escalation_keywords: ["LITELLM ESCALATE"] session_affinity: false complexity_router_default_model: claude-sonnet-5 ``` 如果 Claude Code 或 Claude Desktop 需要發現它，請在 router 名稱中保留 `claude`。請參閱 [管理員設定](/docs/auto_router/setup#claude-code-and-claude-desktop)。 ## OpenAI Family {#openai-family} Luna、Terra、Sol，然後在 xhigh reasoning effort 時使用 Sol。 ```yaml title="config.yaml" model_list: - model_name: gpt-5.6-luna litellm_params: model: openai/gpt-5.6-luna api_key: os.environ/OPENAI_API_KEY - model_name: gpt-5.6-terra litellm_params: model: openai/gpt-5.6-terra api_key: os.environ/OPENAI_API_KEY - model_name: gpt-5.6-sol litellm_params: model: openai/gpt-5.6-sol api_key: os.environ/OPENAI_API_KEY - model_name: gpt-5.6-sol-xhigh litellm_params: model: openai/gpt-5.6-sol api_key: os.environ/OPENAI_API_KEY reasoning_effort: xhigh - model_name: gpt-auto litellm_params: model: auto_router/complexity_router complexity_router_config: tiers: SIMPLE: gpt-5.6-luna MEDIUM: gpt-5.6-terra COMPLEX: gpt-5.6-sol REASONING: gpt-5.6-sol-xhigh classifier_type: heuristic escalation_keywords: ["LITELLM ESCALATE"] session_affinity: false complexity_router_default_model: gpt-5.6-terra ``` ## Gemini Family {#gemini-family} Flash Lite 2.5、Flash Lite 3.1、Flash 3.7，然後是 Pro 3.1。 ```yaml title="config.yaml" keep-model-ids model_list: - model_name: gemini-2.5-flash-lite litellm_params: model: gemini/gemini-2.5-flash-lite api_key: os.environ/GEMINI_API_KEY - model_name: gemini-3.1-flash-lite litellm_params: model: gemini/gemini-3.1-flash-lite api_key: os.environ/GEMINI_API_KEY - model_name: gemini-3.7-flash litellm_params: model: gemini/gemini-3.7-flash api_key: os.environ/GEMINI_API_KEY - model_name: gemini-3.1-pro-preview litellm_params: model: gemini/gemini-3.1-pro-preview api_key: os.environ/GEMINI_API_KEY - model_name: gemini-auto litellm_params: model: auto_router/complexity_router complexity_router_config: tiers: SIMPLE: gemini-2.5-flash-lite MEDIUM: gemini-3.1-flash-lite COMPLEX: gemini-3.7-flash REASONING: gemini-3.1-pro-preview classifier_type: heuristic escalation_keywords: ["LITELLM ESCALATE"] session_affinity: false complexity_router_default_model: gemini-3.1-flash-lite ``` ## Lite {#lite} 跨提供者，以成本為導向。DeepSeek V4 Flash 也可作為 LLM 分類器，採用 `agentic` rubric，並使用零輪次上下文視窗。 ```yaml title="config.yaml" model_list: - model_name: deepseek-v4-flash litellm_params: model: deepseek/deepseek-v4-flash api_key: os.environ/DEEPSEEK_API_KEY - model_name: muse-spark-1.2-xhigh litellm_params: model: meta/muse-spark-1.2 api_key: os.environ/META_API_KEY reasoning_effort: xhigh - model_name: kimi-k3-max litellm_params: model: moonshot/kimi-k3 api_key: os.environ/MOONSHOT_API_KEY reasoning_effort: max - model_name: claude-opus-5 litellm_params: model: anthropic/claude-opus-5 api_key: os.environ/ANTHROPIC_API_KEY - model_name: lite-router litellm_params: model: auto_router/complexity_router complexity_router_config: tiers: SIMPLE: deepseek-v4-flash MEDIUM: muse-spark-1.2-xhigh COMPLEX: kimi-k3-max REASONING: claude-opus-5 classifier_type: llm classifier_llm_config: model: deepseek-v4-flash timeout_ms: 3000 classification_rubric: agentic classifier_context_window_size: 0 escalation_keywords: ["LITELLM ESCALATE"] session_affinity: false complexity_router_default_model: muse-spark-1.2-xhigh ``` ## 1M Context {#1m-context} Luna、Terra、Sol，然後在高 reasoning effort 時使用 Opus 5。使用 heuristic v2 分類器，因此分類不會增加任何 LLM 請求。 GPT 層級最多可接受 922K 輸入 token，而 Opus 5 可接受 1M。router 會將超出大小的提示移到具備容量的下一個最低較高層級。請參閱 [context-window escalation](/docs/proxy/auto_routing#context-window)。 ```yaml title="config.yaml" model_list: - model_name: gpt-5.6-luna litellm_params: model: openai/gpt-5.6-luna api_key: os.environ/OPENAI_API_KEY - model_name: gpt-5.6-terra litellm_params: model: openai/gpt-5.6-terra api_key: os.environ/OPENAI_API_KEY - model_name: gpt-5.6-sol litellm_params: model: openai/gpt-5.6-sol api_key: os.environ/OPENAI_API_KEY - model_name: claude-opus-5-high litellm_params: model: anthropic/claude-opus-5 api_key: os.environ/ANTHROPIC_API_KEY reasoning_effort: high - model_name: 1m-auto litellm_params: model: auto_router/complexity_router complexity_router_config: tiers: SIMPLE: gpt-5.6-luna MEDIUM: gpt-5.6-terra COMPLEX: gpt-5.6-sol REASONING: claude-opus-5-high classifier_type: heuristic_v2 escalation_keywords: ["LITELLM ESCALATE"] session_affinity: false complexity_router_default_model: gpt-5.6-terra ``` ## 基準測試設定 {#the-benchmark-configuration} - [Terminal-Bench](/blog/auto-router-terminal-bench-benchmark)：Opus-5 solve rate 低 27% 成本，這個設定，gpt-5.4-mini 分類器只讀取目前訊息。 - [Cost and quality](/blog/auto-router-cost-quality-benchmark) 與 [prompt caching](/blog/auto-router-prompt-caching-benchmark)：相同層級，heuristic 分類器。 ```yaml title="config.yaml" keep-model-ids model_list: - model_name: claude-haiku-4-5 litellm_params: model: anthropic/claude-haiku-4-5 api_key: os.environ/ANTHROPIC_API_KEY - model_name: claude-sonnet-5 litellm_params: model: anthropic/claude-sonnet-5 api_key: os.environ/ANTHROPIC_API_KEY - model_name: claude-opus-5 litellm_params: model: anthropic/claude-opus-5 api_key: os.environ/ANTHROPIC_API_KEY - model_name: gpt-5.4-mini litellm_params: model: openai/gpt-5.4-mini api_key: os.environ/OPENAI_API_KEY - model_name: smart-router litellm_params: model: auto_router/complexity_router complexity_router_config: tiers: SIMPLE: claude-haiku-4-5 MEDIUM: claude-sonnet-5 COMPLEX: claude-opus-5 REASONING: claude-opus-5 classifier_type: llm classifier_llm_config: model: gpt-5.4-mini classifier_context_window_size: 0 complexity_router_default_model: claude-sonnet-5 ``` 分類器上下文視窗是您自己流量最值得重新檢視的調整旋鈕： - Terminal-Bench：最近 3 則使用者訊息讓 solve rate 從 66.7% 提升到 76.2%，成本增加 44%。assistant 回覆會讓兩者都更差。 - 聊天流量（[v1.97 measurements](/blog/auto-router-context-and-benchmarks)）：先前輪次讓後續回應一致率從 14% 提升到 78%。 - 已發布的預設值：3 則使用者輪次，不含 assistant 輪次。 ## Production config {#the-production-configuration} - 報告 51.1% 節省的 [production case study](/blog/auto-router-production-savings)。 - Haiku 同時處理 SIMPLE 與 MEDIUM；Opus 保留給 REASONING。 - 95% 的請求從未到達旗艦層級。 ```yaml title="config.yaml" keep-model-ids model_list: - model_name: claude-haiku-4-5 litellm_params: model: anthropic/claude-haiku-4-5 api_key: os.environ/ANTHROPIC_API_KEY - model_name: claude-sonnet-5 litellm_params: model: anthropic/claude-sonnet-5 api_key: os.environ/ANTHROPIC_API_KEY - model_name: claude-opus-5 litellm_params: model: anthropic/claude-opus-5 api_key: os.environ/ANTHROPIC_API_KEY - model_name: claude-auto-latest litellm_params: model: auto_router/complexity_router complexity_router_config: tiers: SIMPLE: claude-haiku-4-5 MEDIUM: claude-haiku-4-5 COMPLEX: claude-sonnet-5 REASONING: claude-opus-5 complexity_router_default_model: claude-haiku-4-5 ``` ## 具備負載平衡的 coding agents {#coding-agents-with-load-balancing} - 適用於其後方有多個部署的層級，例如 Anthropic 與 Bedrock 上的相同 Claude model。 - `session_affinity` 會將該層級鎖定於該 session；`deployment_affinity` 加上 `prompt_caching` 的 pre-call 檢查會將具備快取的部署鎖定。 - 這兩者對 agent 流量都很重要。細節：[Session affinity](/docs/proxy/auto_routing#session-affinity)。 ```yaml title="config.yaml" keep-model-ids model_list: - model_name: claude-haiku-4-5 litellm_params: model: anthropic/claude-haiku-4-5 api_key: os.environ/ANTHROPIC_API_KEY - model_name: claude-sonnet-5 litellm_params: model: anthropic/claude-sonnet-5 api_key: os.environ/ANTHROPIC_API_KEY - model_name: claude-sonnet-5 litellm_params: model: bedrock/us.anthropic.claude-sonnet-5 aws_region_name: us-east-1 - model_name: claude-opus-5 litellm_params: model: anthropic/claude-opus-5 api_key: os.environ/ANTHROPIC_API_KEY - model_name: claude-auto litellm_params: model: auto_router/complexity_router cache_control_injection_points: - location: message role: system complexity_router_config: tiers: SIMPLE: claude-haiku-4-5 MEDIUM: claude-sonnet-5 COMPLEX: claude-sonnet-5 REASONING: claude-opus-5 session_affinity: true session_affinity_ttl_seconds: 3600 complexity_router_default_model: claude-sonnet-5 router_settings: optional_pre_call_checks: ["deployment_affinity", "session_affinity", "prompt_caching"] deployment_affinity_ttl_seconds: 3600 ``` ## Related pages - [User Setup](https://docs.litellm.ai/docs/auto_router/user_setup.md) - [Public Benchmarks](https://docs.litellm.ai/docs/auto_router/benchmarks.md)