{"@context":"https://schema.org","@type":"NewsArticle","generatedAt":"2026-07-23T06:40:50.084Z","headline":"OpenRouter 推出 Prompt Caching + Sticky Routing，降低多轮 Agent 调用成本","description":"OpenRouter 通过 Prompt Caching 与 Sticky Routing 降低多轮 Agent 的 token 成本。缓存读取价格仅为正常输入的 0.1x-0.5x，其中 Claude Sonnet 4.6 缓存读取为 $0.30/M（正常 $3.00/M）。","url":"https://www.aioga.com/news/cmruybn7500lbbinvrohd9rio/","mainEntityOfPage":"https://www.aioga.com/news/cmruybn7500lbbinvrohd9rio/","datePublished":"2026-07-21T00:00:00.000Z","dateModified":"2026-07-21T00:00:00.000Z","inLanguage":"zh-CN","publisher":{"@type":"NewsMediaOrganization","name":"Aioga","url":"https://www.aioga.com"},"citation":["https://openrouter.ai/blog/tutorials/prompt-caching-sticky-routing","https://aihot.virxact.com/items/cmruybn7500lbbinvrohd9rio"],"canonicalUrl":"https://www.aioga.com/news/cmruybn7500lbbinvrohd9rio/","directAnswer":{"@type":"Answer","text":"OpenRouter 介绍 Prompt Caching 与 Sticky Routing：重复提示内容可从缓存读取，并通过会话路由回到持有热缓存的同一提供方，从而降低多轮 Agent 的输入 token 成本。","url":"https://www.aioga.com/news/cmruybn7500lbbinvrohd9rio/","dateCreated":"2026-07-21T00:00:00.000Z","author":{"@type":"Organization","@id":"https://www.aioga.com/authors/aioga-editorial/#editorial-team","name":"Aioga Editorial Team","url":"https://www.aioga.com/authors/aioga-editorial/"}},"evidence":[{"@type":"CreativeWork","name":"openrouter.ai source article","url":"https://openrouter.ai/blog/tutorials/prompt-caching-sticky-routing","datePublished":"2026-07-21T00:00:00.000Z","provider":{"@type":"Organization","name":"openrouter.ai","url":"https://openrouter.ai/blog/tutorials/prompt-caching-sticky-routing"}},{"@type":"CreativeWork","name":"AIHot archive record","url":"https://aihot.virxact.com/items/cmruybn7500lbbinvrohd9rio","datePublished":"2026-07-21T00:00:00.000Z","provider":{"@type":"Organization","name":"AIHot","url":"https://aihot.virxact.com/items/cmruybn7500lbbinvrohd9rio"}}],"aggregationSource":"OpenRouter：Announcements（RSS）","originalPublisher":{"name":"openrouter.ai","url":"https://openrouter.ai/blog/tutorials/prompt-caching-sticky-routing"},"article":{"id":"cmruybn7500lbbinvrohd9rio","slug":"cmruybn7500lbbinvrohd9rio","url":"https://www.aioga.com/news/cmruybn7500lbbinvrohd9rio/","title":"OpenRouter 推出 Prompt Caching + Sticky Routing，降低多轮 Agent 调用成本","title_en":"The Cheapest Token Is a Cached One： Prompt Caching + Sticky Routing","summary":"OpenRouter 通过 Prompt Caching 与 Sticky Routing 降低多轮 Agent 的 token 成本。缓存读取价格仅为正常输入的 0.1x-0.5x，其中 Claude Sonnet 4.6 缓存读取为 $0.30/M（正常 $3.00/M）。","source":"OpenRouter：Announcements（RSS）","sourceUrl":"https://openrouter.ai/blog/tutorials/prompt-caching-sticky-routing","aiHotUrl":"https://aihot.virxact.com/items/cmruybn7500lbbinvrohd9rio","publishedAt":"2026-07-21T00:00:00.000Z","category":"技巧观点","score":69,"selected":true,"articleBody":["Your agent sends the same system prompt, tool definitions, schemas, and policy instructions on every turn. In a 6-turn session, you can get billed for that same opening block 6 times, even though the only thing that changed is the user’s latest message or the agent’s latest tool result.","Prompt caching fixes that. The provider reads the repeated part of your prompt from cache instead of charging you full price for it every time. Sticky routing keeps that working across turns by sending a session back to the same provider that holds the warm cache.","This post covers the money side: what cached tokens cost, why cache reads and writes are priced differently, how session_id keeps an agent’s session warm from turn one, and how to check that caching is actually working.","A cache read costs 0.1x to 0.5x of normal input pricing, depending on the provider. That range is why caching can make agent loops much cheaper.","The repeated part is usually the expensive part: a long system prompt, tool definitions, JSON schemas, guardrails, retrieved documents, or examples that keep the model consistent. Without caching, each turn pays full price for all of it again. With caching, the first request writes it to the cache, and later requests read it back at the cheaper rate.","The prompt caching docs：https://openrouter.ai/docs/guides/best-practices/prompt-caching have the full breakdown. The exact dollar amount still depends on the model and provider route; the multiplier tells you how cached input compares with normal input for that provider.","For agent builders, the pattern is simple: the first turn may pay to set up the cache, but every turn after that gets much cheaper as long as the same opening block is reused.","Prompt caching has two costs: the write and the read.","The write happens when the provider stores the reusable part of the prompt. The read happens when a later request reuses that stored content. You come out ahead once the same content gets read enough times to cover the write.","On some providers, the write costs more than normal input. Anthropic cache writes cost 1.25x input for the default 5-minute TTL and 2.0x input for the 1-hour TTL. A single Anthropic cache write that’s never reused costs more than sending the same prompt without caching.","For a one-off request, caching may not help. For a multi-turn agent, repetition is the default: the agent carries the same instructions, tools, schemas, and policy context across the whole session. So the write pays for itself over a few turns.","Use the 5-minute cache lifetime (TTL) for short bursts where the next turn arrives quickly. Use the 1-hour one when a session may pause long enough for the default cache to expire, but the content is still worth keeping around.","A warm cache only helps if the next request lands on the provider endpoint that holds it.","When requests can route to many providers, turn one can write a cache on one provider while turn two lands somewhere else. The second provider has no warm cache to read from. The request still works, but you pay full price and cached_tokens stays low or zero.","That’s why we pair sticky routing with prompt caching. After a cached request, we route follow-up requests for the same model back to the same provider endpoint when that provider’s cache-read pricing is cheaper than normal input. If that sticky provider becomes unavailable, OpenRouter falls back to the next available provider instead of failing the request.","By default, OpenRouter recognizes a conversation by hashing its first system or developer message and its first non-system message. That works when those opening messages stay the same.","Agents often break that. Some rewrite their first messages as they summarize state, reorder tool context, or add new run metadata. When the opening messages change, the hash changes, and the conversation can land on a different provider. The fix is an explicit session_id .","For agent loops, set session_id . When you pass it, OpenRouter uses it directly as the sticky routing key instead of deriving a key from the opening messages.","With session_id , sticky routing kicks in after the first successful request, before any cache hit has happened. Without it, stickiness only starts after a cache hit is detected. For multi-turn agents, that’s the difference between a cache that’s reliable from turn one and one that’s only sometimes warm.","You can send session_id as a top-level request body field or through the x-session-id header. Keep it stable for the conversation or agent run, and keep it under 256 characters.","Use a value that matches the unit of work: a chat thread, ticket, workflow run, or agent task. Don’t create a new session_id for every turn, or requests stop landing on the provider that holds the cache.","If you use router models such as Auto Router：https://openrouter.ai/docs/guides/routing/routers/auto-router#session-stickiness or Pareto Router：https://openrouter.ai/docs/guides/routing/routers/pareto-router#session-stickiness, session stickiness also pins the model the router picked, not just the provider. That keeps the conversation from switching models mid-session, so behavior stays consistent and the cache stays warm.","The fastest way to check is to inspect usage：https://openrouter.ai/docs/cookbook/administration/usage-accounting.","In the response, usage.prompt_tokens_details.cached_tokens shows how many tokens were read from cache. If it’s greater than zero, the request hit the cache. cache_write_tokens shows how many tokens were written during a cache-write request.","In this example, most of the prompt tokens came from cache, and this turn didn’t write a new cache entry.","You can inspect cache behavior in three places: the detail view on the Activity page：https://openrouter.ai/activity, the /api/v1/generation API, and the usage.prompt_tokens_details object returned with API responses.","Use cache_discount to see what a generation saved. On providers with paid writes, you may see a negative discount on the write turn because the cache write costs more than normal input. On later cache-read turns, the discount should turn positive.","When caching looks broken, it usually comes down to one of 4 things: the prompt is too short, the cache expired, the opening content changed, or the request moved to a different provider.","Every provider has a minimum prompt size, and below it nothing gets cached. On Anthropic, Claude Opus 4.5 through 4.8 and Claude Haiku 4.5 need 4,096 tokens; Claude Haiku 3.5 needs 2,048; Claude Sonnet 4, 4.5, and 4.6 (and Opus 4 / 4.1) need 1,024. OpenAI needs 1,024. Gemini 2.5 Pro needs 4,096; Gemini 2.5 Flash needs 1,024.","If your reusable content sits below that minimum, caching won’t start. Don’t pad the request with filler text just to force it. Use caching where you already have a lot of reusable content: tools, schemas, retrieved documents, examples, or policy text.","Caches don’t live long. Anthropic’s default is 5 minutes, with a 1-hour option for longer sessions. Gemini’s implicit cache lasts about 3-5 minutes and doesn’t reset when you read from it. Once a cache expires, the next request has to write a new one.","If your users often pause between turns, use a longer TTL where supported, or build the agent to accept a new write after idle periods.","Automatic and implicit caching work best when the start of the prompt stays the same. Put the stable stuff first: system instructions, tools, schemas, and fixed reference material. Put the changing stuff later: user questions, timestamps, temporary state, tool outputs, and short-lived metadata.","Small details count here. A timestamp in the first system message makes the prompt look new on every turn. Move it into a later user or tool message if it doesn’t need to be part of the cached content.","A cache lives where it was written. If a later request routes to a different provider endpoint, that endpoint can’t read the earlier cache.","For agent workflows, set session_id and let sticky routing keep the session on the warm provider. One catch: if you set provider.order yourself, your order wins over sticky routing. Use provider routing controls：https://openrouter.ai/docs/guides/routing/provider-selection if you need a specific provider order.","If your agent sends the same content every turn, here’s the checklist:","For a rough picture, imagine an agent that repeats the same 10,000 tokens over 6 turns.","This example covers only the repeated content. It ignores the smaller changing messages and the model’s output tokens. The savings grow with the number of turns.","When your agent sends the same expensive content again and again, cached reads and sticky routing keep it from becoming the most expensive part of the loop.","Yes. OpenRouter supports prompt caching across supported providers and models. Most providers enable it automatically, while Anthropic and Alibaba Qwen use cache_control for explicit caching. Cache reads cost 0.1x to 0.5x of normal input pricing depending on the provider, so a reused prefix gets much cheaper after the first request.","Cache reads cost 0.1x to 0.5x of normal input pricing, depending on the provider. Anthropic, DeepSeek, and Alibaba Qwen can read at 0.1x. OpenAI reads at 0.25x to 0.50x. Gemini, Grok, and Moonshot read at 0.25x. Groq reads at 0.5x.","The common causes are a prompt below the provider’s token minimum, an expired cache, an unstable prompt prefix, or provider drift between turns. For agent workflows, start by setting a stable session_id , then check cached_tokens in the usage response, where any value above zero confirms a cache hit.","Pass a stable session_id for the conversation, ticket, or workflow run. OpenRouter uses it as the sticky routing key, so follow-up requests route back to the same provider endpoint that holds the warm cache. With session_id set, stickiness activates after the first successful request, before any cache hit is observed.","Inspect usage.prompt_tokens_details.cached_tokens for cache reads and cache_write_tokens for cache writes; a cached_tokens value above zero confirms a hit. You can also read cache_discount in the response to see the per-generation cost effect, or open the detail view on the Activity page：https://openrouter.ai/activity or the /api/v1/generation API.","Yes. With a session_id set, router models such as Auto Router and Pareto Router pin both the resolved model and the provider for the conversation, so follow-up turns keep hitting the same warm cache."],"articleImages":[{"sourceUrl":"https://openrouter.ai/blog/images/prompt-caching-sticky-routing.png","alt":"The Cheapest Token Is a Cached One: Prompt Caching + Sticky Routing","afterParagraph":0,"url":"/media/articles/cmruybn7500lbbinvrohd9rio/701ee339519bea58.png"},{"sourceUrl":"https://openrouter.ai/blog/images/prompt-caching-agent-session.png","alt":"Diagram of an agent session where session_id keeps turns 1 through N on the same provider for cache writes and cheap cache reads, with automatic failover to a new provider if the sticky one fails","afterParagraph":16,"url":"/media/articles/cmruybn7500lbbinvrohd9rio/771fad506d2eda97.png"}],"mediaStatus":"ok","articleBodyZh":["你的代理每次都会发送相同的系统提示、工具定义、模式和政策指令。在一个6轮的会话中，你可能会因为同样的开头块被计费6次，即使唯一变化的是用户的最新消息或代理的最新工具结果。","提示缓存可以解决这个问题。提供商会从缓存中读取你提示中重复的部分，而不是每次都按全价收费。粘性路由通过将会话发送回拥有热缓存的同一提供商，从而在各轮之间保持这一效果。","本文讨论金钱方面：缓存的令牌成本、为什么缓存读取和写入的价格不同、session_id 如何从第一轮开始保持代理会话的热状态，以及如何检查缓存是否真正有效。","缓存读取的成本为正常输入价格的0.1倍到0.5倍，具体取决于提供商。正因如此，缓存可以显著降低代理循环的成本。","重复的部分通常是昂贵的部分：长系统提示、工具定义、JSON模式、保护措施、检索的文档或示例，这些内容保持模型的一致性。如果没有缓存，每轮都要为所有内容支付全额费用。有了缓存，第一次请求会将它写入缓存，之后的请求以更低的价格读取。","提示缓存文档：https://openrouter.ai/docs/guides/best-practices/prompt-caching 有完整的细分说明。具体金额仍取决于模型和提供商路线；乘数告诉你对于该提供商，缓存输入与正常输入相比如何。","对于代理构建者，这一模式很简单：第一轮可能需要支付建立缓存的费用，但只要复用相同的开头块，之后每轮都将便宜得多。","提示缓存有两种成本：写入和读取。","写入发生在提供商存储提示可复用部分时。读取发生在之后的请求复用已存储内容时。一旦相同内容被读取足够多次来覆盖写入成本，你就会获得收益。","在某些提供商上，写入成本高于正常输入。Anthropic 缓存写入默认5分钟TTL成本为输入量的1.25倍，1小时TTL成本为输入量的2.0倍。一次从未被复用的Anthropic缓存写入，其成本比不使用缓存直接发送相同提示还要高。","对于一次性请求，缓存可能无济于事。对于多轮代理，重复是默认行为：代理在整个会话中携带相同的指令、工具、模式和策略上下文。所以在几轮之后，写入就能收回成本。","对于下一轮很快到来的短时突发，使用5分钟的缓存生命周期（TTL）。当会话可能暂停足够长时间以致默认缓存过期，但内容仍然值得保留时，使用1小时的缓存。","只有当下一次请求落在拥有缓存的提供者端点时，热缓存才有帮助。","当请求可以路由到多个提供者时，第一轮可以在一个提供者上写入缓存，而第二轮落在其他地方。第二个提供者没有可读的热缓存。请求仍然可以执行，但你要支付全价，cached_tokens 保持在低或零。","这就是为什么我们将粘性路由与提示缓存结合使用。在缓存请求之后，我们会将同一模型的后续请求路由回同一提供者端点，当该提供者的缓存读取价格低于正常输入时。如果该粘性提供者不可用，OpenRouter 会退回到下一个可用提供者，而不会导致请求失败。","默认情况下，OpenRouter 通过对话的第一条系统或开发者消息以及第一条非系统消息进行哈希来识别对话。当这些开场消息保持不变时，这种方法有效。","代理经常打破这一点。有些在总结状态、重新排列工具上下文或添加新的运行元数据时，会重写它们的第一条消息。当开场消息变化时，哈希也会变化，导致对话可能落在不同的提供者上。解决方法是显式设置 session_id。","对于代理循环，需要设置 session_id。当你传递它时，OpenRouter 会直接将其用作粘性路由键，而不是从开场消息派生键。","有了 session_id，粘性路由在第一次成功请求后立即生效，而不依赖任何缓存命中。没有它，粘性只在检测到缓存命中后才开始。对于多轮代理，这意味着从第一轮就可靠的缓存与有时才是热缓存的缓存之间的区别。","您可以将 session_id 作为顶层请求体字段发送，或通过 x-session-id 头发送。在会话或代理运行期间保持其稳定，并保持在 256 字符以内。","使用与工作单元匹配的值：聊天线程、工单、工作流运行或代理任务。不要为每个回合创建新的 session_id，否则请求将无法到达持有缓存的提供者。","如果您使用路由模型，例如 Auto Router：https://openrouter.ai/docs/guides/routing/routers/auto-router#session-stickiness 或 Pareto Router：https://openrouter.ai/docs/guides/routing/routers/pareto-router#session-stickiness，则会话粘性不仅固定提供者，还会固定路由器选择的模型。这可防止会话中途中切换模型，从而保持行为一致并保持缓存活跃。","检查的最快方法是查看用量：https://openrouter.ai/docs/cookbook/administration/usage-accounting。","在响应中，usage.prompt_tokens_details.cached_tokens 显示从缓存读取了多少 tokens。如果大于零，表示请求命中了缓存。cache_write_tokens 显示在缓存写入请求期间写入了多少 tokens。","在此示例中，大多数提示 tokens 来自缓存，这一回合没有写入新的缓存条目。","您可以在三个地方检查缓存行为：活动页面上的详细视图：https://openrouter.ai/activity、/api/v1/generation API，以及 API 响应返回的 usage.prompt_tokens_details 对象。","使用 cache_discount 查看生成节省了多少。在付费写入的提供者上，您可能会在写入回合看到负折扣，因为缓存写入比普通输入更昂贵。在后续的缓存读取回合中，折扣应为正数。","当缓存看起来损坏时，通常归结为四种情况之一：提示太短、缓存过期、开头内容更改，或者请求转移到不同的提供者。","每个提供者都有最小提示大小，低于该大小时不进行缓存。在 Anthropic 上，Claude Opus 4.5 到 4.8 和 Claude Haiku 4.5 需要 4,096 个 tokens；Claude Haiku 3.5 需要 2,048；Claude Sonnet 4、4.5 和 4.6（以及 Opus 4 / 4.1）需要 1,024。OpenAI 需要 1,024。Gemini 2.5 Pro 需要 4,096；Gemini 2.5 Flash 需要 1,024。","如果你的可重复使用内容低于这个最小值，缓存将不会开始。不要为了强制缓存而在请求中填充无关文字。只在你已经有大量可重复使用内容的地方使用缓存：工具、模式、检索到的文档、示例或政策文本。","缓存寿命不长。Anthropic 的默认缓存时间是 5 分钟，长会话可选择 1 小时。Gemini 的隐式缓存大约持续 3-5 分钟，读取时不会重置。一旦缓存过期，下一个请求必须写入一个新的缓存。","如果你的用户在轮次之间经常暂停，请在支持的情况下使用更长的 TTL，或构建代理以在空闲期间接受新的写入。","当提示的开头保持不变时，自动和隐式缓存效果最好。将稳定的内容放在前面：系统指令、工具、模式和固定参考资料。将易变的内容放在后面：用户问题、时间戳、临时状态、工具输出和短期元数据。","细节很重要。首条系统消息中的时间戳会使每次提示看起来都是新的。如果不需要成为缓存内容的一部分，请将它移动到后续的用户或工具消息中。","缓存存在于其被写入的位置。如果后续请求路由到不同的提供者端点，该端点无法读取之前的缓存。","对于代理工作流，设置 session_id 并让粘性路由保持会话在热提供者上。有一个注意点：如果你自己设置了 provider.order，你的顺序会优先于粘性路由。如果你需要特定的提供者顺序，请使用提供者路由控制：https://openrouter.ai/docs/guides/routing/provider-selection。","如果你的代理每轮发送相同内容，请检查以下清单：","大致想象一个代理，在 6 轮中重复相同的 10,000 个令牌。","此示例仅涵盖重复内容。它忽略了较小的变化消息和模型的输出令牌。节省量会随着轮数增加而增长。","当你的代理反复发送相同的高成本内容时，缓存读取和粘性路由防止其成为循环中最昂贵的部分。","是的。OpenRouter 支持在支持的提供商和模型之间进行提示缓存。大多数提供商会自动启用，而 Anthropic 和阿里巴巴 Qwen 使用 cache_control 进行显式缓存。缓存读取费用取决于提供商，为正常输入价格的 0.1 倍到 0.5 倍，因此在第一次请求后重用的前缀成本会大大降低。","缓存读取费用取决于提供商，为正常输入价格的 0.1 倍到 0.5 倍。Anthropic、DeepSeek 和阿里巴巴 Qwen 的读取费用为 0.1 倍。OpenAI 的读取费用为 0.25 倍到 0.50 倍。Gemini、Grok 和 Moonshot 的读取费用为 0.25 倍。Groq 的读取费用为 0.5 倍。","常见原因包括提示低于提供商的最小令牌数、缓存过期、提示前缀不稳定，或者轮次之间提供商的漂移。对于代理工作流，首先设置一个稳定的 session_id，然后检查 usage 响应中的 cached_tokens，任何大于零的值都确认命中缓存。","为对话、工单或工作流运行传递一个稳定的 session_id。OpenRouter 使用它作为粘性路由键，因此后续请求会路由回同一个拥有热缓存的提供商端点。设置 session_id 后，粘性在第一次成功请求后激活，即在观察到任何缓存命中之前。","检查 usage.prompt_tokens_details.cached_tokens 以查看缓存读取情况，以及 cache_write_tokens 以查看缓存写入情况；cached_tokens 值大于零表示缓存命中。你也可以在响应中读取 cache_discount 来查看每次生成的成本影响，或者在活动页面打开详细信息视图：https://openrouter.ai/activity 或 /api/v1/generation API。","是的。在设置了 session_id 后，路由模型如 Auto Router 和 Pareto Router 会为对话固定解析模型和提供商，因此后续轮次会持续命中同一个热缓存。"],"translationStatus":"translated","bodyOrigin":"source-page","editorial":{"summary":"OpenRouter 介绍 Prompt Caching 与 Sticky Routing：重复提示内容可从缓存读取，并通过会话路由回到持有热缓存的同一提供方，从而降低多轮 Agent 的输入 token 成本。","background":"多轮 Agent 往往在每轮重复发送系统提示、工具定义、JSON Schema、策略指令或检索文档。材料举例称，六轮会话可能对相同开头区块计费六次，即使变化仅来自新消息或工具结果。","viewpoint":"Aioga 判断，这套方案的关键不只是启用缓存，还在于保持重复区块稳定并维持会话路由。缓存读取价格为正常输入的 0.1 倍至 0.5 倍，但实际金额仍取决于模型和提供方路由。","implications":"对重复上下文占比较高的 Agent，后续轮次可能明显受益；首轮可能承担缓存写入成本，之后才以较低价格读取。若会话未回到持有热缓存的提供方，预期的连续缓存效果可能无法维持。","nextStep":"值得关注的是先识别系统提示、工具定义和 Schema 等稳定区块，再使用 session_id 维持会话热缓存，并检查缓存是否实际生效；成本评估还应分别核对缓存写入、读取及对应提供方价格。","evidenceRefs":["title","summary","articleBody","source"],"status":"published","aiGenerated":true,"autoApproved":true,"generatedBy":"aioga-editorial:gpt-5.6-sol","reviewedBy":"aioga-editorial-review:gpt-5.6-sol","generatedAt":"2026-07-23T01:28:38.387Z","sourceHash":"9e3592c2cbaeeeec","review":{"approved":true,"groundedness":98,"clarity":94,"duplicationRisk":18,"blockingIssues":[],"notes":["候选内容准确区分了来源事实与“Aioga 判断”，未将观点伪装成来源直接结论。","“若会话未回到持有热缓存的提供方，预期的连续缓存效果可能无法维持”属于对 Sticky Routing 作用的合理概括，与来源材料一致。","可选措辞优化：可将“之后才以较低价格读取”改为“后续请求在命中缓存时可按较低价格读取”，以更明确地体现缓存命中这一前提。"]},"validation":{"passed":true,"mode":"ai-auto","revisions":0,"checks":["schema","length","source-attribution","low-source-overlap","no-html","independent-ai-review"]}},"tags":["技巧观点","OpenRouter：Announcements（RSS）"],"translations":{"zh-CN":{"title":"OpenRouter 推出 Prompt Caching + Sticky Routing，降低多轮 Agent 调用成本","summary":"OpenRouter 通过 Prompt Caching 与 Sticky Routing 降低多轮 Agent 的 token 成本。缓存读取价格仅为正常输入的 0.1x-0.5x，其中 Claude Sonnet 4.6 缓存读取为 $0.30/M（正常 $3.00/M）。","category":"技巧观点","source":"openrouter.ai","aggregationSource":"OpenRouter：Announcements（RSS）","pageTitle":"OpenRouter 推出 Prompt Caching + Sticky Routing，降低多轮 Agent 调用成本 - Aioga AI资讯","description":"OpenRouter 通过 Prompt Caching 与 Sticky Routing 降低多轮 Agent 的 token 成本。缓存读取价格仅为正常输入的 0.1x-0.5x，其中 Claude Sonnet 4.6 缓存读取为 $0.30/M（正常 $3.00/M）。","url":"https://www.aioga.com/news/cmruybn7500lbbinvrohd9rio/"},"en":{"title":"OpenRouter launches Prompt Caching + Sticky Routing to reduce the cost of multi-turn Agent calls","summary":"OpenRouter reduces token costs for multi-turn Agents through Prompt Caching and Sticky Routing. The cache read price is only 0.1x-0.5x of normal input, with Claude Sonnet 4.6 cache reads costing $0.30/M (normal $3.00/M).","category":"Insights","source":"OpenRouter：Announcements（RSS）","aggregationSource":"OpenRouter：Announcements（RSS）","pageTitle":"OpenRouter launches Prompt Caching + Sticky Routing to reduce the cost of multi-turn Agent calls - Aioga AI News","description":"OpenRouter reduces token costs for multi-turn Agents through Prompt Caching and Sticky Routing. The cache read price is only 0.1x-0.5x of normal input, with Claude Sonnet 4.6 cache...","url":"https://www.aioga.com/en/news/cmruybn7500lbbinvrohd9rio/","contentTranslated":true,"sourceHash":"640f93d1b08268f8","translatedAt":"2026-07-22T17:25:14.140Z"},"ja":{"title":"OpenRouter、プロンプトキャッシュとスティッキー・ルーティングを導入し、マルチターンエージェント呼び出しのコストを削減","summary":"OpenRouter は、Prompt Caching と Sticky Routing によって複数ラウンドのエージェントのトークンコストを削減します。キャッシュ読み取り価格は通常の入力の 0.1x〜0.5x で、Claude Sonnet 4.6 のキャッシュ読み取りは $0.30/M（通常 $3.00/M）です。","category":"ヒントと視点","source":"OpenRouter：Announcements（RSS）","aggregationSource":"OpenRouter：Announcements（RSS）","pageTitle":"OpenRouter、プロンプトキャッシュとスティッキー・ルーティングを導入し、マルチターンエージェント呼び出しのコストを削減 - Aioga AIニュース","description":"OpenRouter は、Prompt Caching と Sticky Routing によって複数ラウンドのエージェントのトークンコストを削減します。キャッシュ読み取り価格は通常の入力の 0.1x〜0.5x で、Claude Sonnet 4.6 のキャッシュ読み取りは $0.30/M（通常 $3.00/M）です。","url":"https://www.aioga.com/ja/news/cmruybn7500lbbinvrohd9rio/","contentTranslated":true,"sourceHash":"640f93d1b08268f8","translatedAt":"2026-07-22T17:25:34.173Z"},"ko":{"title":"OpenRouter가 프롬프트 캐싱 + 스티키 라우팅을 출시하여 다중 회차 에이전트 호출 비용을 낮춥니다","summary":"OpenRouter는 Prompt Caching과 Sticky Routing을 통해 다회차 Agent의 토큰 비용을 줄입니다. 캐시 읽기 가격은 정상 입력의 0.1x-0.5x에 불과하며, Claude Sonnet 4.6 캐시 읽기는 $0.30/M(정상 $3.00/M)입니다.","category":"인사이트","source":"OpenRouter：Announcements（RSS）","aggregationSource":"OpenRouter：Announcements（RSS）","pageTitle":"OpenRouter가 프롬프트 캐싱 + 스티키 라우팅을 출시하여 다중 회차 에이전트 호출 비용을 낮춥니다 - Aioga AI 뉴스","description":"OpenRouter는 Prompt Caching과 Sticky Routing을 통해 다회차 Agent의 토큰 비용을 줄입니다. 캐시 읽기 가격은 정상 입력의 0.1x-0.5x에 불과하며, Claude Sonnet 4.6 캐시 읽기는 $0.30/M(정상 $3.00/M)입니다.","url":"https://www.aioga.com/ko/news/cmruybn7500lbbinvrohd9rio/","contentTranslated":true,"sourceHash":"640f93d1b08268f8","translatedAt":"2026-07-22T17:26:18.916Z"},"es":{"title":"OpenRouter lanza almacenamiento en caché de prompts + enrutamiento pegajoso, reduciendo el costo de llamadas multironda de agentes","summary":"OpenRouter reduce el costo de tokens de agentes de múltiples rondas mediante almacenamiento en caché de prompts y enrutamiento fijo. El precio de lectura de la caché es solo de 0.1x-0.5x del ingreso normal, donde la lectura de caché de Claude Sonnet 4.6 es de $0.30/M (normal $3.00/M).","category":"Ideas","source":"OpenRouter：Announcements（RSS）","aggregationSource":"OpenRouter：Announcements（RSS）","pageTitle":"OpenRouter lanza almacenamiento en caché de prompts + enrutamiento pegajoso, reduciendo el costo de llamadas multironda de agentes - Aioga Noticias de IA","description":"OpenRouter reduce el costo de tokens de agentes de múltiples rondas mediante almacenamiento en caché de prompts y enrutamiento fijo. El precio de lectura de la caché es solo de 0.1...","url":"https://www.aioga.com/es/news/cmruybn7500lbbinvrohd9rio/","contentTranslated":true,"sourceHash":"640f93d1b08268f8","translatedAt":"2026-07-22T17:26:15.635Z"},"fr":{"title":"OpenRouter lance le caching de prompts et le routage collant, réduisant le coût des appels multi-tours des agents","summary":"OpenRouter réduit le coût en tokens des agents multi-tours grâce à la mise en cache des prompts et au routage collant. Le prix de lecture du cache est seulement de 0,1x à 0,5x du coût normal d'entrée, avec une lecture de cache de Claude Sonnet 4.6 à 0,30 $/M (normalement 3,00 $/M).","category":"Analyses","source":"OpenRouter：Announcements（RSS）","aggregationSource":"OpenRouter：Announcements（RSS）","pageTitle":"OpenRouter lance le caching de prompts et le routage collant, réduisant le coût des appels multi-tours des agents - Aioga Actualités IA","description":"OpenRouter réduit le coût en tokens des agents multi-tours grâce à la mise en cache des prompts et au routage collant. Le prix de lecture du cache est seulement de 0,1x à 0,5x du c...","url":"https://www.aioga.com/fr/news/cmruybn7500lbbinvrohd9rio/","contentTranslated":true,"sourceHash":"640f93d1b08268f8","translatedAt":"2026-07-22T17:27:08.477Z"},"de":{"title":"OpenRouter führt Prompt-Caching + Sticky-Routing ein, um die Kosten für mehrstufige Agent-Aufrufe zu senken","summary":"OpenRouter senkt die Token-Kosten für Multi-Turn-Agenten durch Prompt Caching und Sticky Routing. Der Preis für das Abrufen aus dem Cache beträgt nur 0,1x-0,5x des normalen Eingabepreises, wobei das Abrufen aus dem Claude Sonnet 4.6-Cache $0,30/M kostet (normal $3,00/M).","category":"技巧观点","source":"OpenRouter：Announcements（RSS）","aggregationSource":"OpenRouter：Announcements（RSS）","pageTitle":"OpenRouter führt Prompt-Caching + Sticky-Routing ein, um die Kosten für mehrstufige Agent-Aufrufe zu senken - Aioga KI-News","description":"OpenRouter senkt die Token-Kosten für Multi-Turn-Agenten durch Prompt Caching und Sticky Routing. Der Preis für das Abrufen aus dem Cache beträgt nur 0,1x-0,5x des normalen Eingabe...","url":"https://www.aioga.com/de/news/cmruybn7500lbbinvrohd9rio/","contentTranslated":true,"sourceHash":"640f93d1b08268f8","translatedAt":"2026-07-22T17:27:00.484Z"},"pt-BR":{"title":"OpenRouter lança Cache de Prompt + Roteamento Fixo, reduzindo o custo de chamadas de agente em múltiplas rodadas","summary":"OpenRouter reduz o custo de tokens de agentes de múltiplas rodadas por meio de Prompt Caching e Sticky Routing. O preço de leitura do cache é apenas 0,1x-0,5x do custo de entrada normal, sendo a leitura de cache do Claude Sonnet 4.6 $0,30/M (normal $3,00/M).","category":"技巧观点","source":"OpenRouter：Announcements（RSS）","aggregationSource":"OpenRouter：Announcements（RSS）","pageTitle":"OpenRouter lança Cache de Prompt + Roteamento Fixo, reduzindo o custo de chamadas de agente em múltiplas rodadas - Aioga Notícias de IA","description":"OpenRouter reduz o custo de tokens de agentes de múltiplas rodadas por meio de Prompt Caching e Sticky Routing. O preço de leitura do cache é apenas 0,1x-0,5x do custo de entrada n...","url":"https://www.aioga.com/pt-BR/news/cmruybn7500lbbinvrohd9rio/","contentTranslated":true,"sourceHash":"640f93d1b08268f8","translatedAt":"2026-07-22T17:27:50.338Z"},"ru":{"title":"OpenRouter выпустил Prompt Caching + Sticky Routing, снижая расходы на многораундовые вызовы агенту","summary":"OpenRouter снижает затраты токенов многоходовых агентов с помощью кэширования подсказок и постоянной маршрутизации. Стоимость чтения кэша составляет всего 0,1x-0,5x от обычного ввода, при этом чтение кэша Claude Sonnet 4.6 стоит $0,30/М (обычно $3,00/М).","category":"技巧观点","source":"OpenRouter：Announcements（RSS）","aggregationSource":"OpenRouter：Announcements（RSS）","pageTitle":"OpenRouter выпустил Prompt Caching + Sticky Routing, снижая расходы на многораундовые вызовы агенту - Aioga Новости ИИ","description":"OpenRouter снижает затраты токенов многоходовых агентов с помощью кэширования подсказок и постоянной маршрутизации. Стоимость чтения кэша составляет всего 0,1x-0,5x от обычного вво...","url":"https://www.aioga.com/ru/news/cmruybn7500lbbinvrohd9rio/","contentTranslated":true,"sourceHash":"640f93d1b08268f8","translatedAt":"2026-07-22T17:27:56.104Z"},"ar":{"title":"أطلقت OpenRouter خاصية تخزين التعليمات البرمجية المؤقتة + التوجيه الثابت، لتقليل تكلفة استدعاءات الوكيل متعددة الجولات","summary":"يقلل OpenRouter من تكلفة الرموز لوكلاء المتعدد الجولات عن طريق التخزين المؤقت للمطالبة والتوجيه الثابت. سعر قراءة التخزين المؤقت هو فقط 0.1x-0.5x من الإدخال العادي، حيث أن قراءة التخزين المؤقت لـ Claude Sonnet 4.6 هي $0.30/مليون (العادي $3.00/مليون).","category":"技巧观点","source":"OpenRouter：Announcements（RSS）","aggregationSource":"OpenRouter：Announcements（RSS）","pageTitle":"أطلقت OpenRouter خاصية تخزين التعليمات البرمجية المؤقتة + التوجيه الثابت، لتقليل تكلفة استدعاءات الوكيل متعددة الجولات - Aioga أخبار الذكاء الاصطناعي","description":"يقلل OpenRouter من تكلفة الرموز لوكلاء المتعدد الجولات عن طريق التخزين المؤقت للمطالبة والتوجيه الثابت. سعر قراءة التخزين المؤقت هو فقط 0.1x-0.5x من الإدخال العادي، حيث أن قراءة ال...","url":"https://www.aioga.com/ar/news/cmruybn7500lbbinvrohd9rio/","contentTranslated":true,"sourceHash":"640f93d1b08268f8","translatedAt":"2026-07-22T17:28:43.219Z"},"hi":{"title":"OpenRouter ने प्रॉम्प्ट कैशिंग + स्टिकी राउटिंग लॉन्च किया, जिससे मल्टी-राउंड एजेंट कॉल की लागत कम होती है","summary":"OpenRouter पृांप्ट कैशिंग और स्टिकी राउटिंग के माध्यम से बहु-राउंड एजेंट के टोकन लागत को कम करता है। कैश पढ़ने की कीमत केवल सामान्य इनपुट की 0.1x-0.5x है, जिसमें Claude Sonnet 4.6 कैश पढ़ने की कीमत $0.30/M है (सामान्य $3.00/M)।","category":"技巧观点","source":"OpenRouter：Announcements（RSS）","aggregationSource":"OpenRouter：Announcements（RSS）","pageTitle":"OpenRouter ने प्रॉम्प्ट कैशिंग + स्टिकी राउटिंग लॉन्च किया, जिससे मल्टी-राउंड एजेंट कॉल की लागत कम होती है - Aioga AI समाचार","description":"OpenRouter पृांप्ट कैशिंग और स्टिकी राउटिंग के माध्यम से बहु-राउंड एजेंट के टोकन लागत को कम करता है। कैश पढ़ने की कीमत केवल सामान्य इनपुट की 0.1x-0.5x है, जिसमें Claude Sonnet 4.6...","url":"https://www.aioga.com/hi/news/cmruybn7500lbbinvrohd9rio/","contentTranslated":true,"sourceHash":"640f93d1b08268f8","translatedAt":"2026-07-22T17:28:53.529Z"},"it":{"title":"OpenRouter lancia il Prompt Caching + Sticky Routing, riducendo i costi delle chiamate multi-turn Agent","summary":"OpenRouter riduce il costo dei token per agenti multi-turno attraverso Prompt Caching e Sticky Routing. Il prezzo della lettura della cache è solo 0,1x-0,5x rispetto all'input normale, con Claude Sonnet 4.6 la lettura della cache costa $0,30/M (normale $3,00/M).","category":"技巧观点","source":"OpenRouter：Announcements（RSS）","aggregationSource":"OpenRouter：Announcements（RSS）","pageTitle":"OpenRouter lancia il Prompt Caching + Sticky Routing, riducendo i costi delle chiamate multi-turn Agent - Aioga Notizie IA","description":"OpenRouter riduce il costo dei token per agenti multi-turno attraverso Prompt Caching e Sticky Routing. Il prezzo della lettura della cache è solo 0,1x-0,5x rispetto all'input norm...","url":"https://www.aioga.com/it/news/cmruybn7500lbbinvrohd9rio/","contentTranslated":true,"sourceHash":"640f93d1b08268f8","translatedAt":"2026-07-22T17:29:48.839Z"},"nl":{"title":"OpenRouter lanceert Prompt Caching + Sticky Routing, waardoor de kosten van meerdere agent-aanroepen worden verlaagd","summary":"OpenRouter verlaagt de tokenkosten van multi-turn agents door Prompt Caching en Sticky Routing. De kosten voor het lezen van de cache bedragen slechts 0,1x-0,5x van normale invoer, waarbij het lezen van de Claude Sonnet 4.6-cache $0,30/M kost (normaal $3,00/M).","category":"技巧观点","source":"OpenRouter：Announcements（RSS）","aggregationSource":"OpenRouter：Announcements（RSS）","pageTitle":"OpenRouter lanceert Prompt Caching + Sticky Routing, waardoor de kosten van meerdere agent-aanroepen worden verlaagd - Aioga AI-nieuws","description":"OpenRouter verlaagt de tokenkosten van multi-turn agents door Prompt Caching en Sticky Routing. De kosten voor het lezen van de cache bedragen slechts 0,1x-0,5x van normale invoer,...","url":"https://www.aioga.com/nl/news/cmruybn7500lbbinvrohd9rio/","contentTranslated":true,"sourceHash":"640f93d1b08268f8","translatedAt":"2026-07-22T17:29:38.265Z"},"tr":{"title":"OpenRouter, Çok Turlu Ajan Çağrı Maliyetini Düşürmek İçin İstek Önbellekleme + Sabit Yönlendirme Özelliğini Tanıttı","summary":"OpenRouter, Prompt Caching ve Sticky Routing aracılığıyla çok turlu ajanların token maliyetlerini düşürüyor. Önbellek okuma fiyatı normal girişin yalnızca 0,1x-0,5x'si kadardır; Claude Sonnet 4.6 önbellek okuma ise $0,30/M (normal $3,00/M) olarak fiyatlandırılmıştır.","category":"技巧观点","source":"OpenRouter：Announcements（RSS）","aggregationSource":"OpenRouter：Announcements（RSS）","pageTitle":"OpenRouter, Çok Turlu Ajan Çağrı Maliyetini Düşürmek İçin İstek Önbellekleme + Sabit Yönlendirme Özelliğini Tanıttı - Aioga AI Haberleri","description":"OpenRouter, Prompt Caching ve Sticky Routing aracılığıyla çok turlu ajanların token maliyetlerini düşürüyor. Önbellek okuma fiyatı normal girişin yalnızca 0,1x-0,5x'si kadardır; Cl...","url":"https://www.aioga.com/tr/news/cmruybn7500lbbinvrohd9rio/","contentTranslated":true,"sourceHash":"640f93d1b08268f8","translatedAt":"2026-07-22T17:30:44.783Z"},"vi":{"title":"OpenRouter ra mắt Prompt Caching + Sticky Routing, giảm chi phí gọi nhiều lượt Agent","summary":"OpenRouter thông qua Prompt Caching và Sticky Routing giảm chi phí token cho Agent nhiều vòng. Giá đọc bộ nhớ đệm chỉ bằng 0,1x-0,5x so với đầu vào bình thường, trong đó đọc bộ nhớ đệm của Claude Sonnet 4.6 là $0,30/M (bình thường $3,00/M).","category":"技巧观点","source":"OpenRouter：Announcements（RSS）","aggregationSource":"OpenRouter：Announcements（RSS）","pageTitle":"OpenRouter ra mắt Prompt Caching + Sticky Routing, giảm chi phí gọi nhiều lượt Agent - Tin tức AI Aioga","description":"OpenRouter thông qua Prompt Caching và Sticky Routing giảm chi phí token cho Agent nhiều vòng. Giá đọc bộ nhớ đệm chỉ bằng 0,1x-0,5x so với đầu vào bình thường, trong đó đọc bộ nhớ...","url":"https://www.aioga.com/vi/news/cmruybn7500lbbinvrohd9rio/","contentTranslated":true,"sourceHash":"640f93d1b08268f8","translatedAt":"2026-07-22T17:30:41.462Z"},"id":{"title":"OpenRouter meluncurkan Prompt Caching + Sticky Routing, mengurangi biaya pemanggilan Agent multi-putaran","summary":"OpenRouter mengurangi biaya token Agent multi-putaran melalui Prompt Caching dan Sticky Routing. Harga pembacaan cache hanya 0,1x-0,5x dari input normal, di mana pembacaan cache Claude Sonnet 4.6 adalah $0,30/M (normal $3,00/M).","category":"技巧观点","source":"OpenRouter：Announcements（RSS）","aggregationSource":"OpenRouter：Announcements（RSS）","pageTitle":"OpenRouter meluncurkan Prompt Caching + Sticky Routing, mengurangi biaya pemanggilan Agent multi-putaran - Berita AI Aioga","description":"OpenRouter mengurangi biaya token Agent multi-putaran melalui Prompt Caching dan Sticky Routing. Harga pembacaan cache hanya 0,1x-0,5x dari input normal, di mana pembacaan cache Cl...","url":"https://www.aioga.com/id/news/cmruybn7500lbbinvrohd9rio/","contentTranslated":true,"sourceHash":"640f93d1b08268f8","translatedAt":"2026-07-22T17:31:30.130Z"},"th":{"title":"OpenRouter เปิดตัว Prompt Caching + Sticky Routing เพื่อลดต้นทุนการเรียกใช้งานหลายรอบของ Agent","summary":"OpenRouter ลดต้นทุนโทเค็นของ Agent หลายรอบผ่านการเก็บแคช Prompt และการกำหนดเส้นทางแบบ Sticky ราคาการอ่านแคชเพียง 0.1x-0.5x ของการป้อนข้อมูลปกติ โดย Claude Sonnet 4.6 อ่านแคชที่ $0.30/M (ปกติ $3.00/M)","category":"技巧观点","source":"OpenRouter：Announcements（RSS）","aggregationSource":"OpenRouter：Announcements（RSS）","pageTitle":"OpenRouter เปิดตัว Prompt Caching + Sticky Routing เพื่อลดต้นทุนการเรียกใช้งานหลายรอบของ Agent - ข่าว AI Aioga","description":"OpenRouter ลดต้นทุนโทเค็นของ Agent หลายรอบผ่านการเก็บแคช Prompt และการกำหนดเส้นทางแบบ Sticky ราคาการอ่านแคชเพียง 0.1x-0.5x ของการป้อนข้อมูลปกติ โดย Claude Sonnet 4.6 อ่านแคชที่ $0....","url":"https://www.aioga.com/th/news/cmruybn7500lbbinvrohd9rio/","contentTranslated":true,"sourceHash":"640f93d1b08268f8","translatedAt":"2026-07-22T17:31:37.259Z"},"pl":{"title":"OpenRouter wprowadza buforowanie promptów + stałe trasowanie, obniżając koszty wielokrotnych wywołań agenta","summary":"OpenRouter poprzez Prompt Caching i Sticky Routing zmniejsza koszty tokenów dla wieloetapowego agenta. Cena odczytu z pamięci podręcznej wynosi tylko 0,1x-0,5x normalnego wejścia, przy czym odczyt z pamięci podręcznej Claude Sonnet 4.6 kosztuje 0,30 USD/M (normalnie 3,00 USD/M).","category":"技巧观点","source":"OpenRouter：Announcements（RSS）","aggregationSource":"OpenRouter：Announcements（RSS）","pageTitle":"OpenRouter wprowadza buforowanie promptów + stałe trasowanie, obniżając koszty wielokrotnych wywołań agenta - Aioga Wiadomości AI","description":"OpenRouter poprzez Prompt Caching i Sticky Routing zmniejsza koszty tokenów dla wieloetapowego agenta. Cena odczytu z pamięci podręcznej wynosi tylko 0,1x-0,5x normalnego wejścia,...","url":"https://www.aioga.com/pl/news/cmruybn7500lbbinvrohd9rio/","contentTranslated":true,"sourceHash":"640f93d1b08268f8","translatedAt":"2026-07-22T17:32:23.425Z"}}}}