This guide uses source checks from Sep 2, 2026. Provider and gateway prices can change; preserve the checked date with every forecast.
Why This Topic Matters Now
The Sep 1 keyword report has no large new non-brand breakout. The strongest Tier 1 signal is still the practical brand cluster: `aiwave.live`, `site:aiwave.live`, `aiwave api`, and a few documentation searches. That means today's useful SEO work is not another model ranking. It is a procurement-grade page that tells a buyer how to preserve the live AIWave rate card used in an API trial.
This guide is for teams in the United States, United Kingdom, Canada, Germany, the Netherlands, Japan, and Singapore that evaluate Chinese AI routes through an OpenAI-compatible gateway. It uses live AIWave pricing checked on Sep 2, 2026 and official provider pages checked the same day. The point is a pricing-version drift monitor: a small record that links each request, forecast, and approval note back to the exact rate evidence the team saw.
Source Facts Checked Today
AIWave /api/pricing checked on Sep 2, 2026 returned success=true, 63 records, pricing_version a42d372ccf0b5dd13ecf71203521f9d2, default group ratio 3, VIP group ratio 1, and OpenAI endpoint metadata. Parsed gateway examples before account-group math included DeepSeek V4 Flash at $0.638 input, $1.914 output, and $0.020288 cache-hit input per 1M tokens; DeepSeek V4 Pro at $1.914 input, $5.742 output, and $0.063736 cache-hit input; GLM-5.1 at $2.10 input, $6.60 output, and about $0.680001 cache-hit input; Kimi K3 at $4.50 input, $22.50 output, and $0.90 cache-hit input; and Qwen3.8 Max at about $2.678 input and $8.034 output.
DeepSeek pricing checked on Sep 2, 2026 lists V4 Flash, V4 Pro, and V4 Flash Vision Exp with OpenAI-format and Anthropic-format base URLs, 1M context, 384K maximum output, cache-hit input, cache-miss input, output billing, and concurrency limits of 2500 for Flash routes and 500 for Pro. It also says weekday peak windows are 01:00-04:00 and 06:00-10:00 UTC and that prices may change.
Z.AI pricing checked on Sep 2, 2026 lists GLM-5.3-Flash promotional pricing at $0.075 input, $0.015 cached input, and $0.25 output per 1M tokens until September 9, 2026 UTC+8. It lists GLM-5.3 and GLM-5.2 at $1.40 input, $0.26 cached input, and $4.40 output. QwenCloud pricing checked today documents pay-as-you-go token billing, Batch API at 50% of real-time token rates, context caching, thinking-token output billing, failed-call billing behavior, and built-in tool fees.
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.
| Ledger field | Why it prevents drift | Owner |
|---|---|---|
| pricing_version | Reproduces the AIWave rate card used by the trial | Platform |
| source_checked_at | Separates today evidence from stale rows | Engineering |
| model_name | Prevents silent route substitution | Route owner |
| token_class | Keeps input, output, and cache-hit rows distinct | Finance |
| account_group | Explains default, VIP, or svip math | Admin |
| provider_source_url | Shows which official row was compared | Procurement |
| recheck_by | Forces review before long forecasts | Buyer |
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 json
import urllib.request
req = urllib.request.Request(
"https://aiwave.live/api/pricing",
headers={"Authorization": "Bearer YOUR_API_KEY_HERE"},
)
with urllib.request.urlopen(req, timeout=20) as response:
price_card = json.load(response)
snapshot = {
"checked_at": "2026-09-02",
"pricing_version": price_card.get("pricing_version"),
"records": len(price_card.get("data", [])),
"route_names": [row["model_name"] for row in price_card.get("data", [])[:10]],
}
print(json.dumps(snapshot, indent=2))
Capture the Gateway Card Before the Trial
A trial should begin by saving the gateway card, not by pasting a model name into production. Store pricing_version, checked timestamp, route count, account group, source URL, and the set of route names visible to the account. That snapshot gives engineering a way to reproduce the decision later. It also gives procurement a clean answer when someone asks which pricing table supported the first forecast.
Keep Official Provider Rows Separate
Official provider pages are essential evidence, but they are not the same object as an AIWave gateway row. DeepSeek owns its direct peak/off-peak rows and concurrency limits. Z.AI owns its GLM direct rows and tool prices. QwenCloud owns Batch API and context-cache billing rules. Kimi owns its token and web-search billing guide. AIWave owns the unified route, account group, pricing_version, and usage ledger. Do not collapse those rows into one blended claim.
Record Account Group Without Changing Tokens
AIWave's current operations rule is clear: VIP status applies to all your tokens automatically. A pricing monitor can record the effective account group, but it should not set or modify token groups. The ledger needs to explain whether a forecast used default, VIP, or svip math. It does not need a workflow that asks developers to choose token groups or rotate keys for billing.
Watch Date-Bound Rows
Date-bound rows create the most common drift. DeepSeek says its prices may vary. Z.AI has a GLM-5.3-Flash promotion that ends on September 9, 2026 UTC+8. QwenCloud says model marketplace rows and billing features can vary by model. The monitor should mark each imported provider row with checked date, expiration date when visible, and a recheck date before purchase order or large monthly forecast.
Put Drift Checks in CI
A simple daily or weekly job can fetch the live AIWave pricing card and compare pricing_version, route names, model_ratio, completion_ratio, cache_ratio, and endpoint metadata against the last approved snapshot. The job should not fail every application build when a rate card changes. Instead, it should open a pricing review task, block new procurement forecasts, and tell route owners which rows changed.
Use Internal Links for Brand Searchers
Readers who search AIWave pricing should land near the evidence path: Pricing, Models docs, Chat Completions, Trust, the brand-query evidence pack, and the billing-group guide. Those links turn a branded query into reproducible technical proof.
Procurement Review
A procurement reviewer should ask for the pricing_version, checked date, account group, provider source URL, token class, model name, estimated token mix, and recheck date. Engineering should attach one redacted successful request and one usage object. Finance should preserve the exact calculation sheet. If those artifacts are missing, the team may still run a small test, but it should not approve a long forecast.
Final Checklist
A pricing-version drift monitor is ready when it captures the live gateway card, separates provider rows from gateway rows, records account group, watches date-bound rows, compares changes without modifying production tokens, and links brand searchers to current docs. Recheck AIWave and official provider pages before any procurement decision that depends on exact rates.