Mercury 3 Pricing

Mercury 3 costs $0.20 per million input tokens and $0.70 per million output tokens at regular rates, with prompt caching priced at $0.02 per million cached input tokens. During its launch period as of October 2026, Inception Labs cuts those figures by 50 percent to $0.10 per million input tokens, $0.35 per million output tokens, and $0.01 per million cached tokens.

New developer accounts receive a one-time credit of 100 million free tokens shared across all models on the Inception platform. Inception prices Mercury 3 per million tokens, so spending scales with prompt size, output length, and how much input is cached.

Mercury 3 Price Per Million Tokens

Mercury 3 costs $0.20 per million input tokens and $0.70 per million output tokens at regular rates, and $0.10 and $0.35 at Inception's launch rates as of October 2026. Cached input is $0.02 per million tokens at regular rates and $0.01 at launch, a 90 percent discount on the input price.

Inception Labs announced Mercury 3 on October 19, 2026, presenting it as a diffusion-based large language model (LLM) that produces tokens in parallel. While the docs home page carries a '50% off at launch' banner, Inception Labs publishes no expiration date for the introductory pricing schedule.

A Worked Monthly Cost

Input and output tokens are billed at different rates. According to the Inception Labs documentation frequently asked questions (FAQ), one token represents approximately four characters of English text.

For a baseline of 1,000,000 input tokens and 1,000,000 output tokens without caching, our arithmetic shows a total charge of $0.90 under regular rates ($0.20 plus $0.70). Under the active launch discount, that identical volume totals $0.45 ($0.10 plus $0.35).

To illustrate an active automation pipeline processing 100 million input tokens and 20 million output tokens in a billing cycle, our arithmetic shows the following breakdown depending on caching utilization:

Cached input tokens cost 90 percent less than uncached input: $0.02 per million tokens at regular rates and $0.01 at launch. Each completion returns token counts, but no dollar amounts.

Free Tokens and the Free Tier

Inception Labs provides every new developer account with a one-time grant of 100 million free tokens upon registration (Inception's July 2026 Mercury 2 post describes the credit per new API key). This free balance is pooled across the entire catalog rather than distributed per individual model, meaning calls to mercury-3, mercury-2.5, or mercury-2 draw down the same initial credit.

Inception Labs publishes no recurring free allowance. When the 100 million free tokens run low, add payment information on the platform's billing page to continue.

Free tier usage is subject to per-minute rate limits on requests and tokens.

Rate Limits by Tier

Application programming interface (API) throughput on the Inception platform is regulated through per-minute limits on total requests, input tokens, and output tokens. These limits apply to the Inception API; the Pay As You Go tier has higher limits than Free.

When an application exceeds any of these thresholds within a minute, the endpoint returns an HTTP 429 status code. The Inception documentation advises developers to implement exponential backoff algorithms to manage temporary 429 rate limit and 503 service unavailable responses gracefully.

Organizations that require throughput exceeding 3,000 requests or 3,000,000 input tokens per minute can request quota expansions by contacting Inception support.

Price Differences Across the Mercury Line

Mercury 3 matches Mercury 2.5 on input price ($0.20) and is below both older models on output price ($0.70 against $0.75).

Among the three models in the table above, the 50 percent launch discount applies to Mercury 3 only; Mercury 2.5 and Mercury 2 are shown at list price. Inception's docs also list Mercury Voice at $0.40 input ($0.20 at launch) and $1.50 output ($0.75 at launch), and Mercury Decide at $0.04 regular input ($0.02 at launch).

The guides Mercury 3 Explained and Mercury 3 vs Mercury 2 cover the structural changes between releases, and Mercury 2 API Pricing covers the older rate card.

Third-Party Channels and Dedicated Capacity

Inception does not publish the rates OpenRouter, Baseten or Vercel charge for Mercury 3, and they may differ from the Inception API.

For high-volume production deployments, Inception Labs provides Enterprise agreements featuring dedicated serving capacity, custom data retention rules, and compliance controls. Dedicated capacity pricing is custom (contact Inception) and is not listed on public tables.

Budgeting Checklist and Usage Tracking

Reasoning tokens are billed at the output rate per Inception's FAQ, so the reasoning_effort setting changes your output bill.

Mercury 3 defaults to a low reasoning effort when the parameter is omitted, whereas Mercury 2.5 and Mercury 2 default to medium. Inception recommends high for complex multi-step reasoning.

API completions return token counts in a usage object. Because the response omits calculated dollar fees, financial reporting scripts must multiply recorded tokens by active billing rates.

Teams building long-term financial plans should model expenses under both promotional launch prices and standard rates. If the launch discount ends, plan for at least the regular rates of $0.20 per million input tokens and $0.70 per million output tokens. Inception does not say whether those regular rates will change.

As of October 2026, Mercury 3 fits agent and voice loops with high call volume, at $0.10 and $0.35 per million tokens (Inception's launch rates) and $0.20 and $0.70 at regular rates. Re-check the Inception models table before each month's budget, because the discount has no published end date.

• Read the usage object in API responses to track token counts; no dollar-cost field is returned.