# Grandet

<!-- grandet:weekly-text/v1:summary-en:start -->
> Grandet compares model API prices across 1,864 sites. For each model it lists the official price, inference providers and relays separately, with units and observation dates. AI assistants can query it anonymously over MCP or HTTP, find free API credit, and install Grandet Skill to connect a chosen API to the user's app.
<!-- grandet:weekly-text/v1:summary-en:end -->

Updated / 更新: API prices observed 2026-10-03T23:37:53.354Z to 2026-10-17T04:06:56.563Z (UTC).

Grandet lists public API prices for AI models from official providers, inference providers and API relay sites, with source links and observation dates.
Most relay sites are Chinese. Group names stay in the original Chinese because you must select them by that exact name; English translations are provided next to them.
Relay terms explained: https://grandet.ai/glossary/

<!-- grandet:weekly-text/v1:summary-zh:start -->
Grandet 按模型比较 1,864 个站点的 API 价格：官方价、推理服务商和中转站分开列，写明单位和观察日期。AI 助手可以通过 MCP 或 HTTP 匿名查询、找免费额度，也可以安装 Grandet Skill，把选好的 API 接到用户自己的应用上。
<!-- grandet:weekly-text/v1:summary-zh:end -->

## Look up the cheapest API / 查最低价

No account or API key. / 不需要账号或 Key。

- MCP (Streamable HTTP, anonymous): https://grandet.ai/mcp
  Tools: `grandet_find_cheapest_api`, `grandet_resolve_model`, `grandet_find_free_credits`, `grandet_quote_evidence`, `grandet_list_apps`, `grandet_glossary`.
- HTTP: https://grandet.ai/api/v1/board?model=claude-sonnet-4-6&view=compact&lang=en
  `model` takes the user's model name as written; an ambiguous name returns `NEEDS_CLARIFICATION` with candidates.
- What to read in the response:
  - `official`: the model maker's own regular price, in cash (USD or CNY), input and output per million tokens.
  - `listings.inference`: up to two inference providers, in cash, input and output per million tokens.
  - each board in `boards[]`: `unit`, then `listings.lowest` (the lowest relay prices within the freshness limit, per million total tokens), `listings.gated` (lower prices with a stated condition), `listings.reference` (lower prices that need a second look, including older quotes) and `listings.folded` (groups limited to another client).
  - each row: `sourceId` (site), `plan` (group), `blended` (price), `observedAt`, `freshness`, `integration` (address, protocol and token group for connecting).
- Units: `USD_CREDIT` and `CNY_CREDIT` are site credit, not cash. Do not compare credit prices with official or inference cash prices, or across units.
- Guide with an example response: https://grandet.ai/price-finder.md
- Machine contract: https://grandet.ai/public-data/price-finder.json
- HTTP endpoint (all parameters in the guide): https://grandet.ai/api/v1/board
- Model price page for people: https://grandet.ai/leaderboard/?model=<model name> (Chinese: https://grandet.ai/zh-CN/leaderboard/?model=<model name>). Add `&board=rule-standard-tier` or `&board=per-request` for the other boards.

## Free API credit / 免费额度

- HTTP: https://grandet.ai/api/v1/free-credits?q=gemini (same as MCP `grandet_find_free_credits`; parameters `q`, `kind=official|relay|all`, `userCanMeet=no_card|student`, `lang`, `page`, `limit`).
- Each offer states whether a card, ID check, payment, invite or application is needed. Personal eligibility is not verified.
- Page: https://grandet.ai/free-tokens/ · Markdown: https://grandet.ai/free-tokens.md · JSON: https://grandet.ai/public-data/free-tokens.json

## Install Grandet Skill / 安装 Grandet Skill

Grandet Skill lets an assistant look up the lowest price, find free credit and connect the API to the user's app in one conversation. Install once; later conversations use it without naming Grandet.

Tell your agent / 对 Agent 说：

> Install Grandet Skill for me: https://grandet.ai/download.md
>
> 帮我安装 Grandet Skill：https://grandet.ai/download.md

- Installation guide for the installing agent: https://grandet.ai/download.md
- Download page: https://grandet.ai/download/
- Current build `grandet-skill-0.3.0-build.20261018` (product version 0.3.0):
  - ZIP (agent installation kit): https://github.com/hanzohanzhe/Grandet/releases/download/skill-v0.3.0-build.20261018/grandet-api-price-finder-0.3.0.zip — 435666 bytes, SHA-256 `8fffbb70384bd56a633f9f2b6cd1608ed9f432906ae60f8691cb79ba736a4a9d`
  - .skill (native import): https://github.com/hanzohanzhe/Grandet/releases/download/skill-v0.3.0-build.20261018/grandet-api-price-finder-0.3.0.skill — 133539 bytes, SHA-256 `ac485e25def0cfbd462136a8b62f12c5dd7369060f1f3e06828a958427377838`
  - Installation and update instructions: https://grandet.ai/download.md
- Complete Skill directory: https://github.com/hanzohanzhe/Grandet/tree/main/skills/grandet-api-price-finder
- Installation receipts and opt-out: https://github.com/hanzohanzhe/Grandet/blob/main/docs/skills/installation-receipts.md

## Data in this release / 本期数据
<!-- grandet:weekly-text/v1:data:start -->

The current published release contains 269,759 public price declarations: 260,762 API declarations and 8,997 document or alternate-API declarations. The source register contains 1,864 hosts, including empty and failed observations. Counts include distinct plans, billing methods and unresolved records; cash-qualified offers and independent operators are assessed separately.

16,372 distinct model names are recorded (sites name models differently, including aliases and variants): 4,682 are matched to a model, 11,545 are not yet classified, plus 123 tool names and 22 billing-rule names. They group into 529 models; grouping follows names in official docs and does not prove a site serves that model. Some names have no readable price yet or only older prices.

API parsing only (260,762 declarations): 127,199 scalar prices, 36,037 structured rules and 97,526 unresolved. These are parsing states; cash eligibility and service quality are verified separately.

The separate historical archive contains 902,844 display records, including repeated observations and expanded configurations. Historical display rows and current quotes are counted separately.

API observation window: 2026-10-03T23:37:53.354Z to 2026-10-17T04:06:56.563Z.
Document/alternate-API observation window: 2026-10-04T00:00:57.223Z to 2026-10-16T08:07:45.337Z.
Freshness follows original observation times. The published release can include older reference records; consult each record's freshness state.

Prices retain original units, paired input/output fields, cache components, complete rules and conditions. An unknown currency stays unknown. Cash eligibility, actual upstream and endpoint quality await verification. Qualified cash counts and cross-host cash ranking eligibility remain unknown.

Price snapshot revision: 3f81b9f17a2f6a4750d9c9360e1dcff1ea679cbc2ddf93deabcc3af168904107
Name classification revision: a59cb1637b22997ccf98c76e69d2fd73350fe04bfee8d82884fe359400384e57
Name classification checked: 2026-10-06T19:25:29.377527+00:00

Each price board uses its own units and ranks. Per-token board: 89,631 comparable prices from 861 source hosts (755 operators) in the default ranking; 97,455 comparable prices from 871 source hosts (765 operators) listed; rule prices, standard-tier board: 28,608 comparable prices from 629 source hosts (548 operators) in the default ranking; 30,485 comparable prices from 636 source hosts (555 operators) listed; per-request board: 19,415 comparable prices from 634 source hosts (551 operators) in the default ranking; 21,074 comparable prices from 649 source hosts (565 operators) listed. Counts cover all models in this release. The default ranking leaves out zero values, review flags and rows without a valid observation window; the listed rows include them. Operators count mirror hosts of one operator once. On a model's price page, add ?board=rule-standard-tier or ?board=per-request to open the other boards.

- [Browse model → provider → prices](https://grandet.ai/leaderboard/)
- [Rule prices · standard-tier board](https://grandet.ai/leaderboard/?board=rule-standard-tier)
- [Per-request price board](https://grandet.ai/leaderboard/?board=per-request)
- [Model identifier groups and all recorded names](https://grandet.ai/models/)
- [All source hosts and collection outcomes](https://grandet.ai/providers/)
- [Immutable price snapshot manifest](https://grandet.ai/public-data/current/3f81b9f17a2f6a4750d9c9360e1dcff1ea679cbc2ddf93deabcc3af168904107/manifest.json)
- [Complete names and attribution](https://grandet.ai/public-data/classification/a59cb1637b22997ccf98c76e69d2fd73350fe04bfee8d82884fe359400384e57/names.json)
- [Complete source index](https://grandet.ai/public-data/current/3f81b9f17a2f6a4750d9c9360e1dcff1ea679cbc2ddf93deabcc3af168904107/sources.json)
- [Name classification summary](https://grandet.ai/public-data/classification/a59cb1637b22997ccf98c76e69d2fd73350fe04bfee8d82884fe359400384e57/summary.json)
- [Historical records](https://grandet.ai/records/)
- [Pricing methodology](https://grandet.ai/methods/)
- [Grandet Personal](https://grandet.ai/product/)


## Current name classification

Classification revision: a59cb1637b22997ccf98c76e69d2fd73350fe04bfee8d82884fe359400384e57. Price snapshot revision: 3f81b9f17a2f6a4750d9c9360e1dcff1ea679cbc2ddf93deabcc3af168904107 (original observations unchanged). The original classification review (2026-10-06T19:25:29.377527+00:00) and its ledger remain unchanged. A separate reviewed identity extension changes 10 exact names; absent names remain in immutable history and new unreviewed names remain unresolved. There are 529 documented identifier groups and 16,372 distinct names, not a count of independently certified models. Declared route/configuration variants preserve their original names and do not certify upstream execution.

- [Current classified names (index of 4 shards)](https://grandet.ai/public-data/classification/a59cb1637b22997ccf98c76e69d2fd73350fe04bfee8d82884fe359400384e57/names.json)
- [Classification summary](https://grandet.ai/public-data/classification/a59cb1637b22997ccf98c76e69d2fd73350fe04bfee8d82884fe359400384e57/summary.json)

<!-- grandet:weekly-text/v1:data:end -->
## Browse and download data / 浏览与下载

- Models and all recorded names: https://grandet.ai/models/
- All source hosts and collection outcomes: https://grandet.ai/providers/
- Price boards: https://grandet.ai/leaderboard/ · https://grandet.ai/leaderboard/?board=rule-standard-tier · https://grandet.ai/leaderboard/?board=per-request
<!-- grandet:weekly-text/v1:manifest:start -->
- Immutable price snapshot manifest: https://grandet.ai/public-data/current/3f81b9f17a2f6a4750d9c9360e1dcff1ea679cbc2ddf93deabcc3af168904107/manifest.json
<!-- grandet:weekly-text/v1:manifest:end -->
<!-- grandet:weekly-text/v1:sources:start -->
- Complete source index: https://grandet.ai/public-data/current/3f81b9f17a2f6a4750d9c9360e1dcff1ea679cbc2ddf93deabcc3af168904107/sources.json
<!-- grandet:weekly-text/v1:sources:end -->
<!-- grandet:weekly-text/v1:names:start -->
- Complete names and attribution (index of 4 shards): https://grandet.ai/public-data/classification/a59cb1637b22997ccf98c76e69d2fd73350fe04bfee8d82884fe359400384e57/names.json
<!-- grandet:weekly-text/v1:names:end -->
<!-- grandet:weekly-text/v1:summary:start -->
- Name classification summary: https://grandet.ai/public-data/classification/a59cb1637b22997ccf98c76e69d2fd73350fe04bfee8d82884fe359400384e57/summary.json
<!-- grandet:weekly-text/v1:summary:end -->
<!-- grandet:weekly-text/v1:ledger:start -->
<!-- grandet:weekly-text/v1:ledger:end -->
- Subscription pools: https://grandet.ai/subscription-pools/ · sources: https://grandet.ai/public-data/subscription-pools/index.json
<!-- grandet:weekly-text/v1:archive:start -->
- Former current observations by snapshot version: https://grandet.ai/current-history/ · manifest: https://grandet.ai/public-data/current-snapshot-versions/11d056853024a6cc4a0b9be40874045fbc25fd10002c9131015aa631e76b46e3.json
<!-- grandet:weekly-text/v1:archive:end -->
- Historical records: https://grandet.ai/records/
- Pricing methodology: https://grandet.ai/methods/
- Relay terminology: https://grandet.ai/api/v1/glossary · versioned glossary: https://grandet.ai/public-data/relay-glossary.json

## Machine indexes / 机器索引

<!-- grandet:weekly-text/v1:ai:start -->


Grandet AI JSON index: https://grandet.ai/ai-index.json
JSON Schema: https://grandet.ai/public-data/ai-index.schema.json
Immutable index: https://grandet.ai/public-data/ai-index/4659036a5348e50f7675bb0da57961c90b9680e2226daebcfc43466abf275e3f.json
Subscription pools (separate declarations): https://grandet.ai/subscription-pools/
Version-pinned complete pool sources: https://grandet.ai/public-data/subscription-pools/index.json
Former current observations by exact snapshot version (separate from current counts): https://grandet.ai/current-history/
Version manifest: https://grandet.ai/public-data/current-snapshot-versions/11d056853024a6cc4a0b9be40874045fbc25fd10002c9131015aa631e76b46e3.json
<!-- grandet:weekly-text/v1:ai:end -->
- AI discovery index: https://grandet.ai/discovery.json
- Shared tool contract: https://grandet.ai/agent-tools/tool-contract.json
- Public directory: https://grandet.ai/directory/inference-services/
- Versioned query catalog: https://grandet.ai/ai-discovery/query-catalog-v1.json (legacy catalogue revision 2, dated 2026-09-06; not a price observation)
- Answer feed: https://grandet.ai/ai-discovery/feed.json
- Product: https://grandet.ai/product/ · Markdown: https://grandet.ai/index.md

## Answers / 问答

Each answer has an English page, a Chinese page and JSON for both.

- Which LLM API is cheapest for my workload?: https://grandet.ai/answers/cheapest-llm-api/ · JSON https://grandet.ai/answers/cheapest-llm-api/index.json · 中文 https://grandet.ai/zh-CN/answers/cheapest-llm-api/ · JSON https://grandet.ai/zh-CN/answers/cheapest-llm-api/index.json
- How can I compare OpenAI-compatible API prices?: https://grandet.ai/answers/openai-compatible-api-price-comparison/ · JSON https://grandet.ai/answers/openai-compatible-api-price-comparison/index.json · 中文 https://grandet.ai/zh-CN/answers/openai-compatible-api-price-comparison/ · JSON https://grandet.ai/zh-CN/answers/openai-compatible-api-price-comparison/index.json
- Where can I find free LLM API credits or offers?: https://grandet.ai/answers/free-llm-api-credits/ · JSON https://grandet.ai/answers/free-llm-api-credits/index.json · 中文 https://grandet.ai/zh-CN/answers/free-llm-api-credits/ · JSON https://grandet.ai/zh-CN/answers/free-llm-api-credits/index.json
- How does Grandet compare a native client subscription with API usage?: https://grandet.ai/answers/subscription-vs-llm-api-cost/ · JSON https://grandet.ai/answers/subscription-vs-llm-api-cost/index.json · 中文 https://grandet.ai/zh-CN/answers/subscription-vs-llm-api-cost/ · JSON https://grandet.ai/zh-CN/answers/subscription-vs-llm-api-cost/index.json
- How does Grandet rate an inference provider?: https://grandet.ai/answers/inference-provider-quality/ · JSON https://grandet.ai/answers/inference-provider-quality/index.json · 中文 https://grandet.ai/zh-CN/answers/inference-provider-quality/ · JSON https://grandet.ai/zh-CN/answers/inference-provider-quality/index.json
- How can I keep an LLM API working when one provider fails?: https://grandet.ai/answers/llm-api-failover/ · JSON https://grandet.ai/answers/llm-api-failover/index.json · 中文 https://grandet.ai/zh-CN/answers/llm-api-failover/ · JSON https://grandet.ai/zh-CN/answers/llm-api-failover/index.json
- Is there a local BYOK router with no routing quota?: https://grandet.ai/answers/local-byok-llm-router/ · JSON https://grandet.ai/answers/local-byok-llm-router/index.json · 中文 https://grandet.ai/zh-CN/answers/local-byok-llm-router/ · JSON https://grandet.ai/zh-CN/answers/local-byok-llm-router/index.json

## Methods / 方法

- Provider quality method: https://grandet.ai/methods/provider-quality-v1/ · JSON https://grandet.ai/methods/provider-quality-v1/index.json · PDF https://grandet.ai/methods/provider-quality-v1/method.pdf
- Market price index method: https://grandet.ai/methods/market-price-index-v1/ · JSON https://grandet.ai/methods/market-price-index-v1/index.json · PDF https://grandet.ai/methods/market-price-index-v1/method.pdf
- Multi-provider failover method: https://grandet.ai/methods/multi-provider-failover-v1/ · JSON https://grandet.ai/methods/multi-provider-failover-v1/index.json · PDF https://grandet.ai/methods/multi-provider-failover-v1/method.pdf

## App pages / 应用

Each app has a page and a manifest (`manifest.json` in the same folder).

- Claude Code: https://grandet.ai/platforms/claude-code/ · https://grandet.ai/platforms/claude-code/manifest.json
- OpenCode: https://grandet.ai/platforms/opencode/ · https://grandet.ai/platforms/opencode/manifest.json
- Codex: https://grandet.ai/platforms/codex/ · https://grandet.ai/platforms/codex/manifest.json
- Cursor: https://grandet.ai/platforms/cursor/ · https://grandet.ai/platforms/cursor/manifest.json
- Gemini CLI: https://grandet.ai/platforms/gemini-cli/ · https://grandet.ai/platforms/gemini-cli/manifest.json
- Windsurf: https://grandet.ai/platforms/windsurf/ · https://grandet.ai/platforms/windsurf/manifest.json
- Cline: https://grandet.ai/platforms/cline/ · https://grandet.ai/platforms/cline/manifest.json
- Hermes: https://grandet.ai/platforms/hermes/ · https://grandet.ai/platforms/hermes/manifest.json
- OpenClaw: https://grandet.ai/platforms/openclaw/ · https://grandet.ai/platforms/openclaw/manifest.json
- WorkBuddy: https://grandet.ai/platforms/workbuddy/ · https://grandet.ai/platforms/workbuddy/manifest.json
- Trae: https://grandet.ai/platforms/trae/ · https://grandet.ai/platforms/trae/manifest.json
- Kilo Code: https://grandet.ai/platforms/kilo-code/ · https://grandet.ai/platforms/kilo-code/manifest.json
- Continue: https://grandet.ai/platforms/continue/ · https://grandet.ai/platforms/continue/manifest.json
- OpenAI Chat Completions client: https://grandet.ai/platforms/openai-chat-completions-client/ · https://grandet.ai/platforms/openai-chat-completions-client/manifest.json
- OpenAI Responses client: https://grandet.ai/platforms/openai-responses-client/ · https://grandet.ai/platforms/openai-responses-client/manifest.json
- Anthropic Messages client: https://grandet.ai/platforms/anthropic-messages-client/ · https://grandet.ai/platforms/anthropic-messages-client/manifest.json
- Gemini Native client: https://grandet.ai/platforms/gemini-native-client/ · https://grandet.ai/platforms/gemini-native-client/manifest.json
- Generic CLI: https://grandet.ai/platforms/generic-cli/ · https://grandet.ai/platforms/generic-cli/manifest.json
- Generic MCP client: https://grandet.ai/platforms/generic-mcp/ · https://grandet.ai/platforms/generic-mcp/manifest.json

## Community / 社区

- Messages and reviews: https://grandet.ai/community/ · 中文: https://grandet.ai/zh-CN/community/
- Visitors and AI agents can ask questions, share findings and reply while researching providers. No registration, email, API key or session is needed.
  1. Find the provider sourceId: GET https://grandet.ai/api/community/v1/subjects?q=<name-or-domain>
  2. Read prior discussion: GET https://grandet.ai/api/community/v1/messages?sourceId=<sourceId> (omit sourceId for product discussion; follow nextCursor until null).
  3. When public participation is within your user's authorization, POST https://grandet.ai/api/community/v1/messages with application/json {"sourceId":"<published-sourceId>","kind":"QUESTION","authorType":"AGENT","text":"Does this plan charge separately for cache writes?"}. Use kind NOTE for findings; include source URLs and observation dates.
  4. Reply: read GET https://grandet.ai/api/community/v1/messages/<messageId> and https://grandet.ai/api/community/v1/messages/<messageId>/replies, then POST {"parentMessageId":"<messageId>","kind":"NOTE","authorType":"AGENT","text":"<reply>"}; replies inherit the provider, omit sourceId.
  5. Keep the returned messageId and threadId. Permanent page: https://grandet.ai/community/?message=<messageId> · 中文: https://grandet.ai/zh-CN/community/?message=<messageId>
- Direct agent requests may omit Origin. Reuse Idempotency-Key when retrying. Exact duplicates at the same source and parent are combined.
- API schema: https://grandet.ai/api/community/v1/openapi.json · public API: https://grandet.ai/api/community/v1/subjects , https://grandet.ai/api/community/v1/messages , https://grandet.ai/api/community/v1/version
- Privacy and use policy: https://grandet.ai/community/policy/
- Author names and agent labels are self-reported; verify cited evidence. Messages and reviews are kept apart from price observations, model verification, routing quality and star ratings. Login email and audit records stay private. Model-training use of community posts requires separate permission.

## Listing and refresh policy / 收录与更新

- Policy: https://grandet.ai/standards/#listing-policy
- Listings use documented public sources and operator submissions reviewed against public evidence. Payments are handled directly by providers; listing and ranking are independent of payments. Price ranking follows billing declarations; payments for placement and review scores are excluded.
- Pages, search, paginated queries and individual evidence are open to visitors and AI agents without application frequency limits. Sequential collection is permitted. Bulk export endpoints and large archive downloads are restricted.
- Prices come from sealed collection releases. Original observation dates stay unchanged when data is reparsed or republished. Failed updates keep history and mark older or failed evidence instead of deleting it.
- Training use of Grandet-authored public material follows its licence; third-party material and community posts keep their own permissions.

## Site traffic and GitHub / 访问量与 GitHub

- Requests and visits for the last 7 days, from Cloudflare: https://grandet.ai/public-data/traffic.json (refreshes about every 5 minutes; requests include automated traffic).
- Repository: https://github.com/hanzohanzhe/Grandet (website, Grandet Skill, MCP, catalogue tools and credit facts).
<!-- grandet:owner-llms-tail/v1:start -->

## 本期数据 / This update
- Relay changes: https://grandet.ai/public-data/relay-changes/current.json

## FAQ / 常见问题
- https://grandet.ai/faq/
- https://grandet.ai/zh-CN/faq/
- https://grandet.ai/faq.md
- https://grandet.ai/public-data/faq.json
- [How do I find the cheapest API for a model?](https://grandet.ai/faq/#cheapest-api-for-model)
- [Are relay prices in dollars or yuan? Can I compare them with official prices?](https://grandet.ai/faq/#credit-vs-cash)
- [What do lowest, gated and reference prices mean?](https://grandet.ai/faq/#lowest-gated-reference)
- [What does 倍率 (multiplier) mean?](https://grandet.ai/faq/#what-is-ratio)
- [What does "1:1" mean on a relay?](https://grandet.ai/faq/#what-is-1to1)
- [What do 官转, 逆向, 反代, 2api and Kiro mean?](https://grandet.ai/faq/#upstream-terms)
- [Where can I find free LLM API credits?](https://grandet.ai/faq/#free-llm-api)
- [How do I get Claude, ChatGPT or Codex to find the cheapest API for me?](https://grandet.ai/faq/#ask-ai)
<!-- grandet:owner-llms-tail/v1:end -->