Snowflake Cortex pricing

About

Snowflake Cortex is the AI layer of Snowflake’s data cloud — a set of in-database AI capabilities you call straight from SQL or Python without moving data out of the warehouse. It spans AI Functions (AI_COMPLETE, AI_EMBED, AI_CLASSIFY, AI_EXTRACT, plus the earlier SUMMARIZE/TRANSLATE/SENTIMENT/EXTRACT_ANSWER in the SNOWFLAKE.CORTEX schema), model APIs that route to hosted models, the Cortex REST API, and a set of higher-level products: Cortex Agents and Snowflake Intelligence (agentic orchestration), Cortex Analyst (natural-language-to-SQL over your semantic models), Cortex Search / Cortex Batch Search (managed hybrid vector + keyword retrieval for RAG), and AI Parse Doc (document parsing).

Snowflake Inc. (NYSE: SNOW) is public, with over 7,200 brands on the platform and multi-billion-dollar product revenue. Cortex is still not a standalone product with its own subscription — but as of the 2026 pricing refresh it is no longer billed as ordinary Snowflake credits either. Snowflake now separates its AI features onto a dedicated AI Credit meter that is priced independently of your Snowflake edition, while warehouse compute, storage, and every other workload continue to draw down edition-priced Platform Credits. There are still no seats, no Cortex subscription, and no separate invoice — but the AI and non-AI halves of your bill now run on two distinct credit currencies with two distinct pricing rules.

For the live rate card, see Snowflake’s pricing page and the AI pricing docs.

Pricing summary : How Snowflake Cortex’s pricing model works

Cortex is pure consumption, but as of 2026 it runs on two distinct credit currencies, and most AI queries still stack two metered layers:

1. AI Credits (the AI meter) — every AI feature is billed in AI Credits, a flat, edition-independent credit whose dollar price is set by cross-region routing, not by your Snowflake edition: $2.00 per AI Credit for global routing (CORTEX_ENABLED_CROSS_REGION = ANY_REGION / *_GLOBAL) and $2.20 per AI Credit for regional routing (routing disabled or region-locked, for data residency). AI Functions, the Cortex REST API, Cortex Agents, Snowflake Intelligence, Cortex Search / Batch Search, and AI Parse Doc all bill in AI Credits per the Consumption Table; the per-1M-token, per-page, or per-message rate varies by model and function. Automatic ACV-based discounts apply to AI Credits, but pre-paid capacity discounts do not.
2. Platform Credits (the compute meter) — the virtual warehouse that executes the SQL, plus storage and all non-AI usage, draws down Platform Credits at your edition’s per-credit price: $2.00 (Standard), $3.00 (Enterprise), $4.00 (Business Critical), with Virtual Private Snowflake (VPS) sales-quoted. Snowflake still recommends staying at Medium or smaller warehouses for Cortex calls.

The headline change is that the AI half of your bill is no longer edition-multiplied: identical AI usage now costs the same $2.00/AI-Credit on a Business Critical account as on Standard — only the Platform-Credit compute underneath it still varies by edition.

What makes this different: Cortex still doesn’t quote a dollar-per-token rate the way OpenAI or Anthropic do — it sells credits — but it now runs a dedicated AI currency whose price is decoupled from edition and instead keyed to routing/data-residency. Adoption stays frictionless (it’s in your existing contract; no amendment required), but reasoning about a bill means tracking two separate credit types with two separate pricing rules.

Pricing by product

The two credit currencies

Snowflake’s own worked example: an Enterprise account where Platform Credits are $3.00 and AI Credits are $2.00 (global routing), using 100 Platform Credits and 100 AI Credits, spends 100 × $3.00 = $300 (platform) + 100 × $2.00 = $200 (AI) = $500 total.

AI features billed in AI Credits

Note: warehouse compute still applies on top of AI Credits for AI Functions — the warehouse that runs the SQL is billed in Platform Credits by the second (X-Small = 1, doubling per size step to Large = 8 credit-hours). Snowflake recommends staying at Medium or smaller for Cortex calls. Media inputs are tokenized: audio at 50 tokens per second, images by the model, and document pages at 970 tokens each for AI_EXTRACT.

Sales motions across products: Cortex is self-serve / PLG for any existing Snowflake customer (turn it on in SQL, no procurement, no contract amendment for AI Credits), while net-new platform contracts and capacity commitments are sales-led.

Hidden costs : What Snowflake Cortex users actually pay

The “AI Credits per 1M tokens” rate is the part people budget for — and the part that’s almost never the whole bill.

• Two credit currencies, not one. AI features bill in AI Credits; the warehouse that runs the SQL, plus storage, bills in Platform Credits. Your monthly Cortex spend is the sum of both meters, quoted in two different per-credit prices.
• Warehouse time is the silent second meter. Every AI Function call still runs inside a warehouse you’re already paying for by the second, in Platform Credits. Run an AI_COMPLETE across a large table and the warehouse stays hot for the whole scan; the compute credits can dwarf the token credits. Practitioners have reported five-figure single-query surprise bills from exactly this pattern.
• Routing, not edition, sets the AI price. AI Credits cost $2.00 (global routing) or $2.20 (regional routing) regardless of edition — but pinning workloads to a home region for data residency raises the AI-Credit price ~10%. Meanwhile the Platform-Credit compute underneath still varies by edition ($2/$3/$4), so a regulated Business Critical account still pays a 2x premium on the warehouse half of every AI query.
• Capacity discounts don’t reach AI Credits. Pre-paid capacity discounts apply to Platform Credits and storage but explicitly not to AI Credits — only the automatic ACV-based discount reduces the AI meter.
• Cortex Search is always-on. Its serving compute (credits/GB-month) accrues continuously while the index is resumed — you pay to keep it warm even on a day with zero searches. Cortex Batch Search bills credits/GB-hour only while the batch job runs, which better fits intermittent workloads.

The exact per-1M-token, per-page, and per-GB AI-Credit rates are model- and function-specific and are published only in Table 6 of the Snowflake Service Consumption Table (a downloadable document), so a real bill combines the AI-Credit line for tokens with the Platform-Credit line for the warehouse seconds that ran them.

Want to estimate your own Snowflake Cortex bill? Use the Snowflake Cortex pricing calculator to model your costs based on usage patterns.

Pricing evolution : Snowflake Cortex pricing history and changes

Cadence

Tracked range: 2024–present. The 2026 refresh is the first structural change: Snowflake split AI onto a separate AI Credit meter priced by cross-region routing rather than edition, while Platform Credits (warehouse, storage) stay at $2/$3/$4 by edition.

Notable changes

• 2024-05 — Cortex LLM functions reached GA with published per-1M-token credit rates; no new SKU.
• 2025-01 — Cortex Search introduced its continuous serving charge (credits/GB-month); Cortex Analyst bundled at the standard credit rate.
• 2026-07 — Snowflake introduced AI Credit pricing: AI features (AI Functions, Cortex Search / Batch Search, Agents, Snowflake Intelligence, REST API, AI Parse Doc) now bill in a dedicated, edition-independent AI Credit priced by routing — $2.00 global routing / $2.20 regional routing — with automatic ACV-based discounts (capacity discounts excluded). Platform Credit prices ($2/$3/$4 by edition) and $23/TB storage are unchanged. Cortex Fine-tuning and the standalone Cortex Analyst API remain on legacy Platform-Credit pricing.

What’s unique : Snowflake Cortex’s distinctive pricing mechanics

1. A dedicated AI currency, decoupled from edition. As of 2026 Cortex bills AI in AI Credits — a separate credit type whose flat price ($2.00 global / $2.20 regional routing) is the same regardless of Snowflake edition, while every other workload keeps drawing edition-priced Platform Credits. Snowflake is one of very few vendors to run two parallel credit currencies with two different pricing rules on the same bill.

2. Price keyed to routing / data residency, not model or edition. The single lever that moves the AI-Credit price is CORTEX_ENABLED_CROSS_REGION: let requests route globally and you pay $2.00; pin them to a home region for compliance and you pay $2.20. Data-residency posture, not tier, sets the AI unit price.

3. Two stacked meters per query. Almost every other LLM provider bills you once (tokens). Cortex bills the model tokens in AI Credits and the warehouse seconds that ran the SQL in Platform Credits. The compute layer is frequently the larger of the two, which is unique among AI APIs — and it still varies by edition even though the AI half no longer does.

Strengths & weaknesses

Billing UX : Snowflake Cortex billing controls and transparency

• Billing controls — Cortex spend is governed by the same levers as the rest of Snowflake: resource monitors with credit quotas and auto-suspend, warehouse auto-suspend/auto-resume, and budgets. There’s no Cortex-specific cap, so guardrails are warehouse- and account-level.
• Usage visibility — Per-function token and credit consumption is observable via ACCOUNT_USAGE views: CORTEX_FUNCTIONS_USAGE_HISTORY and CORTEX_FUNCTIONS_QUERY_USAGE_HISTORY (per-query, grouped by model), CORTEX_AGENT_USAGE_HISTORY and SNOWFLAKE_INTELLIGENCE_USAGE_HISTORY (per-agent), plus METERING_HISTORY. This lets you attribute credits to specific models and queries — strong transparency once you know where to look, but it requires SQL, not a dashboard tile.
• Two credit currencies on one bill — Since the 2026 refresh, statements separate AI Credits (edition-independent, routing-priced) from Platform Credits (edition-priced compute/storage). No contract amendment is required to start consuming AI Credits, and existing capacity balances are unchanged.
• Payment options — On-demand (pay month-to-month by credit) or pre-paid capacity commitments that unlock volume discounts on Platform Credits and storage — but capacity discounts do not apply to AI Credits (only the automatic ACV-based discount does). Billing flows through the existing Snowflake account; Cortex never invoices separately.

Strategic wins : Why Snowflake Cortex’s pricing decisions worked

1. Frictionless adoption via the existing meter

By refusing to create a separate Cortex SKU, Snowflake made AI a one-line SELECT away for every existing customer — no new contract, no new budget approval. The credit model is the distribution channel. See usage-based pricing strategy for why removing procurement friction accelerates adoption.

2. Data-gravity lock-in

Cortex runs where the data already lives, so the pricing story (“no egress, no data movement, governance intact”) is also a retention story. Customers spend AI credits because moving data out to a standalone model API would be slower and riskier. Related: how AI companies structure pricing and outcome-based pricing trends.

3. Pricing stability through model churn

Snowflake keeps the AI Credit price fixed (by routing, not model) and absorbs model economics into per-token AI-Credit rates in the consumption table. The 2026 move to a dedicated AI Credit sharpens this: customers get a stable unit of account decoupled from both their edition and the volatile underlying model market. See choosing the right usage metric.

Areas to improve : Gaps in Snowflake Cortex’s pricing approach

1. The double meter breeds bill shock

Because warehouse compute is billed alongside tokens, users routinely underestimate Cortex cost — a single AI_COMPLETE over a large table can run a warehouse hot for the whole scan and produce reported five-figure surprise charges. See bill shock and cost unpredictability.

2. Dollar cost is one conversion removed

Quoting AI in credits, not dollars, means every team still has to multiply by the AI-Credit price ($2.00 or $2.20 by routing) — and separately by their edition’s Platform-Credit rate for the warehouse compute — to know what a query actually costs. A native per-function dollar estimator in the UI would close the gap.

3. Always-on Cortex Search charge

Paying serving compute per GB-month just to keep a Cortex Search index resumed penalizes spiky or low-query-volume RAG workloads. The 2026 addition of Cortex Batch Search (billed per GB-hour only while a job runs) partly addresses this, but a true serverless / pay-per-query interactive Search tier would better fit intermittent use.

Key takeaways

1. Cortex now runs two credit currencies. AI features bill in edition-independent AI Credits ($2.00 global / $2.20 regional routing); warehouse compute and storage bill in edition-priced Platform Credits ($2/$3/$4).
2. Budget for two meters, not one. Every AI Function query costs model tokens in AI Credits plus warehouse seconds in Platform Credits — and compute often wins.
3. Routing, not edition, sets the AI price. Decoupling the AI Credit from edition means identical AI usage costs the same across tiers; the only lever is global vs regional (data-residency) routing.
4. Cortex Search is an always-on cost (serving compute per GB-month while resumed), while Cortex Batch Search bills per GB-hour only while a job runs.
5. For the data-platform category, consumption-on-the-warehouse is the winning AI monetization — zero friction, strong lock-in, and now a dedicated AI currency that survives model-market churn.

UBP implications

1. Meter what you already sell. Snowflake bolted AI onto an existing credit meter instead of inventing a new unit — the fastest path to monetizing AI when you already have a consumption relationship.
2. Beware stacked meters. Bundling a hidden second meter (compute) under a headline meter (tokens) maximizes revenue but erodes predictability; expose both clearly or expect bill-shock churn.
3. Decouple a volatile-cost meter from unrelated packaging tiers, and give it a stable unit of account. Snowflake’s 2026 split does both at once: it lifts AI onto its own currency whose price it holds fixed while repricing models underneath (a steady mental model as COGS moves), and it stops gating that AI cost on a slow-moving tier lever (edition) customers bought for unrelated reasons — pricing it instead on a dimension they can actually choose (cross-region routing / data residency). That removes the “regulated accounts pay 2x for identical AI” distortion without touching edition economics — a powerful pattern for any AI reseller whose AI cost has drifted onto the wrong meter.

Sources

• Snowflake pricing — editions & credits (accessed 2026-07-06)
• Snowflake AI (Cortex) pricing docs (accessed 2026-07-06)
• Cost considerations for Cortex AI Functions (accessed 2026-07-06)
• Understanding cost for Cortex Search Services (accessed 2026-07-06)
• Snowflake Service Consumption Table (per-model AI Credit rates, Table 6) (accessed 2026-07-06)

Bottom line

Snowflake Cortex is the cleanest example in the corpus of “AI on the warehouse” pricing — but the 2026 refresh split that warehouse in two. AI features now bill in a dedicated, edition-independent AI Credit priced by cross-region routing ($2.00 global / $2.20 regional), while warehouse compute and storage keep drawing edition-priced Platform Credits ($2 Standard / $3 Enterprise / $4 Business Critical). Adoption stays nearly frictionless for existing customers (no contract amendment for AI Credits), but real Cortex cost now lives across two currencies: AI Credits for the tokens/pages/messages of every AI feature, plus Platform Credits for the warehouse seconds that ran them and always-on charges like Cortex Search’s serving compute. Budget for both meters, watch your warehouse sizing, and remember that capacity discounts reach only the Platform-Credit half.

Want to compare Snowflake Cortex against other data-platform and AI-infrastructure companies? Browse the pricing blueprint.

Pricing timeline : Major events on a vertical axis

Each milestone below corresponds to a public pricing change, product launch, or material adjustment. Major events use a filled marker; minor adjustments use a faded one.

AI Credit pricing introduced — AI split off the edition-priced meter

Jul 2026

Snowflake moved AI features (AI Functions, Cortex Search / Batch Search, Cortex Agents, Snowflake Intelligence, Cortex REST API, AI Parse Doc) onto a dedicated, edition-independent AI Credit priced by cross-region routing — $2.00 global / $2.20 regional. Platform Credits ($2/$3/$4 by edition) and $23/TB storage unchanged; capacity discounts do not apply to AI Credits, only automatic ACV-based discounts.

Consumption table refreshed with frontier + long-context models

Jun 2026

Per-1M-token credit rates expanded to cover newer models (Claude 4.5 family, GPT-5.5) and long-context variants priced at roughly 2x their standard counterparts; edition credit prices held at $2/$3/$4.

Cortex Search and Cortex Analyst pricing formalized

Jan 2025

Cortex Search introduced a continuous serving charge (credits per GB-month of indexed data) plus embedding-token credits; Cortex Analyst bundled at the standard edition credit rate with no separate fee.

Cortex LLM functions reach general availability

May 2024

Snowflake GA'd Cortex LLM functions (COMPLETE, SUMMARIZE, TRANSLATE, SENTIMENT, EXTRACT_ANSWER) with per-1M-token credit pricing published in the Service Consumption Table — no new SKU, billed as standard credits.