Source directory

Model list (models, listing qwen/deepseek/kimi/glm/minimax, etc.)/model pricing (model-pricing)/product updates (release-notes) are all 200 publicly accessible. The console model plaza requires login. QwenLM is the official open-source organization of Qwen. The documentation site also provides API online debugging.

Model list (rmh4stp0j)/model service billing (wmh4sv6ya)/update dynamics (Mmh8l4qwj)/model update records (Kmh4stnjp) are all 200. The document is JS-rendered, so static scraping can only obtain the framework; it is recommended to use a browser for rendering or the official SDK. The console model plaza requires login. Select baidu's parent company (Xueren/PaddlePaddle, etc., open source) under github.

The official pricing page (quick_start/pricing) lists all models (such as deepseek-chat/deepseek-reasoner/V3.2) at the same time, combining model listing and pricing functions. The news page is the changelog. The platform.deepseek.com platform console requires login. The documentation site is publicly accessible and can be crawled directly, supporting /models API.

Model Overview (models-intro) and Pay-as-you-go (pricing-paygo, including language/voice/video/image pricing) are both 200. There is no independent changelog page, so the official website's news (www.minimaxi.com/news) is used as a substitute. The documentation provides llms.txt, and API is compatible with the OpenAI/Anthropic specification.

Important: Kimi open platform documentation has been migrated from platform.moonshot.cn to platform.kimi.com (the old domain remains accessible, with 200). The model parameter reference (api/models-overview), model inference pricing explanation (pricing/chat), and platform new feature release notes (changelog) are the current official pages. The documentation provides llms.txt. github uses the MoonshotAI official organization.

The model plaza cloud.siliconflow.cn/models is public and does not require login; it can be filtered by category and displays the unit price of each model, making it a model list and price data source. Update announcements release-notes/overview are the official changelog. The real path of the documentation is /docs/... (no cn prefix); it provides llms.txt and /v1/models is compatible with API and can be directly fetched.

Product Overview 104753 (including model list)/Hunyuan Text Generation Billing Overview 97731 (including input/output/cache pricing)/Product Announcements 132069 (also API Update History 101849) are 200. api_docs takes the Hunyuan Large Model API overview 101848. Note that the tokenhub.cloud.tencent.com domain is not accessible, and the actual address of the TokenHub console is console.cloud.tencent.com/tokenhub/ (login required). Tencent-Hunyuan is the official Hunyuan open-source organization.

Model list (1330310)/model service pricing (1544106, including input/output/cache)/product update announcements (1159177, with separate model announcement 1159178) are all 200 publicly accessible. The console model plaza requires login. docs 82379 is the document center of the Ark large model service platform.

The platform for React SPA has direct scraping from page URL failing (only returning an empty shell). The update log is available in the public announcement API (JSON, including new model updates, feature iterations, and billing changes), while the model square /modelSquare and pricing page are CSR, which require browser rendering and are temporarily not to be scraped. The dynamic updates in Coding Plan / Token Plan are also included in announcement API.

Model overview (docs.bigmodel.cn/cn/guide/start/model-overview)/update log (new-releases)/product pricing (bigmodel.cn/pricing) are all 200. open.bigmodel.cn/pricing can also serve as the model+pricing page. zai-org is the official open-source organization (Zhipu/Z.ai) of GLM; THUDM is Tsinghua-affiliated. The overseas site is z.ai. The console requires login.

Model Overview (models/overview, including each ID and pricing comparisons)/API Pricing (docs pricing)/API Release Notes, with all 200 public. API/ here is the developer pricing (enterprise/commercial plans are at www.anthropic.com/pricing). docs provides llms.txt for scraping.

Model documentation (docs/models), pricing (cohere.com/pricing), Changelog (docs.cohere.com/changelog), and 200 are all publicly available. txt.cohere.com is the official technical blog. The dashboard.cohere.com console requires login. docs provides llms.txt for scraping.

Note: ai.google.dev redirects curl without cookie for 302 to Google login, but the page content is actually publicly readable (web_fetch has confirmed Models/Pricing content, including input/output/cache pricing across Gemini 3.x tiers). It and Vertex AI are two separate systems. AI Studio(aistudio.google.com) can be trialed and accessed for API Key for free.

The model page (learn/models)/pricing (pricing, enterprise, includes third-party models such as Llama/GLM/Qwen and cache pricing)/Release Notes are 200 public. Model Garden(console.cloud.google.com/vertex-ai/publishers) requires login. This is the enterprise entry point for MaaS on Google Cloud.

Model overview (models_overview) / official website pricing (mistral.ai/pricing) / documentation API are all 200. Mistral has no standalone changelog page (/updates on docs has been 404), replaced by the official news page. console.mistral.ai is the La Plateforme console (307 → log in).

platform.openai.com/docs/models, /docs/pricing (including input/output/cache pricing), and /docs/changelog are all 200 public. openai.com/news/ returns 403 (anti-scraping) for curl, but it is accessible normally in a browser. The root domain of platform.openai.com requires login (console). /v1/models API is provided.

docs.x.ai/docs/models is titled 'Grok Models & Pricing', with 1 pages simultaneously covering the model list and pricing (including input/output/cache prices), and 200 is public. The x.ai main site and /news return 403 for curl (strict anti-crawling), accessible via browsers; for automated scraping, docs.x.ai or GitHub is recommended. xai-org is the official open-source organization.