One AI model can be sold by multiple platforms. This record prices the Mistral-hosted GLM endpoint only, not every Z.ai plan.
Evidence scope: First-party documentation was checked on 25 October 2026. Published pricing is a scoped list reference, not a guaranteed account quote. Source dates, release dates and the first effective date of a price are not assumed to be interchangeable; geography, plan, currency and invoicing may alter the paid amount.
The contract is with the serving platform
Mistral’s 22 October 2026 changelog says zai-glm-5-3-flash became available on its platform at $0.15 input, $0.03 cached input and $0.50 output per million tokens. This is primary evidence for the Mistral-hosted SKU, because Mistral sets that platform’s invoice. It is not a primary Z.ai direct-billing quote. Keep biller=Mistral and underlying model supplier=Z.ai as distinct fields to avoid a misleading “universal GLM rate”.
There are more than one GLM price tiers
The live Mistral USD table also presents GLM 5.3 at $1.40 input, $0.14 cached input and $4.40 output per million, separate from the Flash version. Treat these as two product model IDs: zai-glm-5-3 and zai-glm-5-3-flash. An article that says “GLM costs fifty cents” without specifying output token, Flash and host has made three errors at once.
An illustrative hosted workload
On Mistral’s Flash tariff, two million ordinary input tokens, 300,000 cached input tokens and 700,000 output tokens imply 2×$0.15 + 0.3×$0.03 + 0.7×$0.50 = $0.659. Repeat the calculation with its non-Flash model and the result changes substantially. The illustration excludes retries and does not assert that cached input exists for your application. A buyer must inspect the returned usage counters.
Why region and access status need direct checks
GLM availability at Mistral does not prove that a user can pay Z.ai directly from the same geography. A reseller may expose only selected model versions, currencies, caching behaviours or priorities. The source’s model identifier and October release note are deliberately preserved; if a reseller removes an endpoint or reorders offerings, the existing history should stay intact. Do not infer mainland-China payment support from this Mistral-hosted USD entry.
How to cross-reference providers
The next editorial task for this family is independent first-party Z.ai subscription and direct API pricing, including Chinese domestic versus international endpoints. Until source-specific verification is complete, those direct rates remain unquoted. This article offers a useful hosted example without turning a missing figure into zero or borrowing a rival marketplace’s invoice as fact. Capability, privacy and latency comparisons belong to the model and developer-tool teams.
How to verify the actual invoice
Begin with a small representative workload rather than the cheapest published headline. Record the exact provider and billing channel, API model identifier, currency, account geography, plan and commitment length. The price or quota evidence in this guide is keyed to cached_input_tokens_per_1m, input_tokens_per_1m, output_tokens_per_1m. Do not combine proprietary creative credits, token meters or media seconds into a single currency or invent a conversion rate between them. If an enterprise or regional rate is not published, leave its price unknown and obtain a written quote.
Keep separate estimates for failed attempts, optional tools, storage, exports, human quality review and tax. Use source-verified line items from the companion CSV instead of typing permanent numbers into your forecasting formulas. Where an old plan or model was superseded, preserve its original historical record and create an additive later snapshot. This guide does not claim a product test, performance ranking, independent legal assessment or live procurement approval.
Availability, alternatives and editorial limits
Available plans, supported endpoints, trial eligibility, quota reset and contract rights may vary by region and account even when a public website displays a single global currency. Compare an alternative only with the same accepted task, input/output volume, model version and privacy requirements. Vendor-provided marketing claims are not independent evidence of quality. This historical pricing article may be cross-referenced by the model, media or enterprise procurement team, but its purpose is to explain the billing mechanics without duplicating their evaluations or inventing test scores.
Pricing evidence boundary
This cost discussion concerns Zhipu / Mistral (zai-glm-5-3 / zai-glm-5-3-flash). Reproduce a quote only with its exact model or plan identifier, chargeable unit, billing region, processing mode and timestamp; a changing vendor rate card is not a permanent price guarantee. If any one of these dimensions is missing, keep the associated rate and worked scenario under editorial review instead of filling the gap with an estimate.
Sources & Further Reading
- https://docs.mistral.ai/inference/pricing (official source; checked 2026-10-25)
- https://docs.mistral.ai/resources/changelogs (official source; checked 2026-10-25)