Official Price Page Drift Capture for Chinese AI API Workbooks

Why This Topic Matters Now

The Sep 14 keyword report repeatedly says the same quiet but important thing: official model pricing pages are dynamic, and exact rows should be rechecked before publication. That matters because Chinese AI API workbooks often mix DeepSeek, QwenCloud, Z.AI, Kimi, and gateway rows in one sheet. Without a drift capture, a team may update one provider tab, leave another stale, and then wonder why the monthly forecast no longer matches receipts.

A drift capture is a small source-dated record. It stores the URL, checked date, visible currency, visible unit, copied rows, pricing version when the endpoint exposes one, feature family, reviewer, and downstream workbook tabs that changed. It does not need customer prompts or secrets. It exists so engineering, finance, and procurement can agree on which pricing facts were current when a route was approved.

Source Facts Checked Today

AIWave /api/pricing checked on Sep 15, 2026 returned success=true, 64 live rows, top-level pricing_version a42d372ccf0b5dd13ecf71203521f9d2, auto_groups=['default'], and group_ratio default=1 and vip=0.9. The static /api/v1/pricing endpoint checked during the same run reported checked=2026-09-10, currency=USD, unit=per_1m_text_tokens, pricing_version 8c7a0c0b30661ccbc13d142cb54d1e4ae445fe774b2c6fa501080db97c7a3e56, and notes that dated base rates are adjusted by the effective account group. Selected static rows were moonshot-v1-128k at $1.80 input and $4.50 output per 1M tokens, moonshot-v1-auto at $75 input and $75 output, kimi-k3 at $4.50 input, $0.90 cache-hit input, and $22.50 output, GLM-5.1 at $2.10 input, $0.680001 cache-hit input, and $6.5999997 output, and DeepSeek Flash at $0.70 input, $0.0233 cache-hit input, and $2.10 output. VIP-key estimates multiply the same base rows by 0.9.

QwenCloud pricing opened for this run says billing varies by model type: text per token, image generation per image, video per second, and speech per character or audio second. It also describes Batch API, context caching, thinking-token billing, and built-in tools such as Web Search priced at $10 per 1K calls. Those features should be captured as separate workbook families.

Z.AI pricing opened for this run lists prices in USD, per-1M-token rows for latest and text models, built-in Web Search at $0.01 per use, GLM-Image at $0.015 per image, CogView-4 at $0.01 per image, video model rows per video, audio rows, and agent rows. Kimi pricing opened for this run lists Web Search at $0.004 per invocation and explains context caching for repeated context content.

Planning Matrix

A source-dated planning matrix keeps the page useful for engineers and procurement reviewers. It turns a search query into an auditable route decision instead of a loose model preference.

Implementation Pattern

The implementation pattern keeps credentials as placeholders, pins the AIWave base URL, records the model, and leaves room for route-specific controls. Production applications should move credentials into environment or secret storage.

import requests
from datetime import date

SOURCES = [
    "https://aiwave.live/api/pricing",
    "https://aiwave.live/api/v1/pricing",
    "https://api-docs.deepseek.com/quick_start/pricing/",
    "https://docs.qwencloud.com/developer-guides/getting-started/pricing",
    "https://docs.z.ai/guides/overview/pricing",
    "https://www.kimi.ai/help/kimi-api/api-pricing",
]

for url in SOURCES:
    headers = {"User-Agent": "AIWave price drift capture", "X-API-Key": "YOUR_API_KEY_HERE"}
    response = requests.get(url, timeout=25, headers=headers)
    print({"url": url, "checked": str(date.today()), "status": response.status_code})

Capture Before Copying Rows

A budget workbook should never receive copied prices without a capture record. First capture the page or JSON endpoint, then copy rows into the workbook, then record which tabs changed. If a provider page exposes a pricing version, store it. If it only exposes visible rows, store the checked date and the row labels. The capture should be small enough that teams actually use it.

Separate Units by Feature Family

Text tokens, cached input, tool calls, images, video seconds, audio seconds, speech characters, and agent rows are not interchangeable. QwenCloud and Z.AI both make this visible in their pricing pages. A workbook should therefore separate feature-family tabs instead of averaging every cost into a single Chinese AI API rate. Separate tabs make product gating and incident review much easier.

Keep Direct Providers and Gateway Rows Apart

Direct-provider rows explain what a vendor charges on its own platform. Gateway rows explain what the AIWave route table and receipts show for a request through AIWave. Both are useful, but they answer different questions. Store provider rows in comparison tabs and AIWave rows in gateway tabs. When a route is approved, link the two only through an explicit review note.

Track Group Ratio Separately

AIWave's live table currently shows default=1 and vip=0.9. That multiplier is not a model row; it is a key-group or account-context field that changes the applied estimate. Keep base rates and applied estimates in separate workbook columns. This prevents a later reviewer from mistaking a VIP-key estimate for a public base row.

Add Reviewer Ownership

A drift capture should include a human or automation owner. The owner is responsible for saying whether a source changed enough to update public content, internal calculators, or procurement notes. If no owner is assigned, pricing drift becomes everybody's problem and nobody's checklist. Ownership also reduces accidental public claims based on stale screenshots.

Internal Links for Budget Teams

This article should link to Pricing JSON, Pricing, Models docs, Chat Completions, Status, and Trust. The purpose is to move searchers from price-page drift into current AIWave evidence quickly.

Publication Review

Before a public article includes exact prices, the reviewer should see source URL, checked date, currency, unit, exact row labels, and whether the row is direct provider or AIWave gateway. If an official page mentions temporary rows or limited conditions, avoid turning that into evergreen copy. If the page includes feature-specific billing, do not reduce it to a text-token example without a caveat.

Final Checklist

A price-page drift capture is ready when every source has status, date, unit, copied rows, feature family, owner, and downstream workbook tab. Re-run the capture before changing public pricing content, before funding a production workload, and whenever a pricing_version changes. The habit is small; the avoided confusion is large.