72 / 100

Modal pricing teardown

https://modal.com/pricing

Modal's pricing page is a well-structured usage-based model that leads with granular per-second resource costs before presenting three plan tiers, making it highly transparent for developer-centric buyers. The page does strong work on trust and objection handling (FAQ, serverless vs. traditional cost comparison, marketplace integrations) but lacks annual billing options and visual hierarchy cues like a recommended plan badge. Overall it's a solid self-serve developer page with a few gaps that prevent a top score.

Tier structure

Starter

$0 + compute/mo

Team

$250 + compute/mo

Enterprise

Custom

Value metric

usage-based compute (per CPU core/sec, per GiB/sec memory, per GPU/sec) plus flat platform fee per tier

How limits scale

Workspace seats Up to 3 → Unlimited → Unlimited
Containers 100 → 1000 → Custom
GPU concurrency 10 → 50 → Custom
Log retention 1 day → 30 days → Custom
Deployed apps 200 → 1000 → 1000

Escalation logic

Starter is a free-entry tier with $30 included compute credits and hard limits on seats, containers, and GPU concurrency, designed to onboard individual developers and small teams. Team adds a $250 platform fee, unlocks unlimited seats and cron jobs, raises container and GPU concurrency limits significantly, and adds enterprise-adjacent features like custom domains and deployment rollbacks, while Enterprise is fully custom for security, compliance, and volume needs.

Psychological anchors

  • Usage transparency as trust anchor — Leading with a full per-second GPU/CPU/memory rate card before showing plan tiers signals radical pricing honesty, reducing sticker shock and building credibility with technical buyers.
  • Included compute credits as entry anchor — Starter includes $30/mo free compute and Team includes $100/mo free compute, framing the platform fee as partially offset and lowering perceived cost of upgrading.
  • High-priced enterprise tier as anchor — The 'Custom' Enterprise tier with volume discounts, embedded ML engineering, and HIPAA makes the $250 Team plan feel accessible and well-scoped by comparison.
  • Serverless cost comparison as ROI anchor — A concrete example ($5,400 traditional vs. $4,740 Modal for a GPU workload) is used to justify the usage-based model and reframe price as savings rather than cost.
  • Credit grants as acquisition anchor — Separate credit grant programs for startups and academics ($10k for researchers) act as zero-friction entry points that bypass the Starter plan's limits entirely.
  • Marketplace committed spend — AWS and GCP marketplace transact options let enterprise buyers use pre-committed cloud budgets, reducing procurement friction for large deals.

What this page is optimizing for

This page is primarily optimized for self-serve developer conversion — granular public pricing, instant 'Get Started' CTAs, and a free Starter tier all reduce friction for individual developers and small teams to sign up without talking to sales. A secondary enterprise lead-gen motion exists via 'Get in touch' and private Slack support, but it's clearly subordinate to the bottom-up PLG funnel.

Red flags

  • No annual billing option is offered or even mentioned, leaving a standard 17-20% discount lever unused and potentially losing budget-conscious buyers who prefer annual commitments.
  • No recommended or 'most popular' plan is highlighted, missing a key decoy-effect anchor that would guide undecided buyers toward Team.
  • The 'Deployed apps' limit is identical for Team and Enterprise (both 1000), which weakens the Enterprise upgrade trigger on that dimension and may confuse buyers comparing tiers.
  • The dual pricing table (once in plan cards, once in the full feature comparison grid) creates redundancy and page length without adding clarity — consolidating would reduce cognitive load.
  • Charm pricing ($249 or $299) is not used; the round $250 Team price is fine for premium positioning but misses a psychological nudge that could improve conversion at the margin.
  • No visible social proof (customer logos, testimonials, usage stats) appears in the pricing section itself, which is a missed trust signal at the moment of purchase decision.
  • The Sandbox + Notebooks pricing section uses different CPU/memory rates than the main section without a clear explanation of why, which could confuse buyers and erode trust.

Best-practices scorecard

What works · 4

  • Visible prices — All prices including per-second GPU/CPU rates are fully public and prominently displayed, which is exemplary for a usage-based model.
  • Value metric matches usage — Per-second billing for actual compute consumed is a near-perfect value metric for ML infrastructure workloads that are inherently bursty and variable.
  • FAQ or objection handling — A dedicated FAQ section addresses billing mechanics, credit requirements, seat limits, and marketplace credits — covering the most common pre-purchase objections.
  • Clear CTAs per tier — Each tier has a distinct, action-oriented CTA: 'Get Started' for Starter, 'Sign in to upgrade' for Team, and 'Get in touch' for Enterprise, all appropriately differentiated.

Half measures · 2

  • Tier differences are scannable — The full feature comparison table is comprehensive but long; the plan cards above it are scannable, though the duplication adds friction rather than clarity.
  • Trust signals present — SOC 2, HIPAA, and Okta SSO are mentioned in the Enterprise tier and comparison table, but no customer logos or testimonials appear in the pricing section.

What's missing · 2

  • Clear recommended tier — No plan is visually or textually marked as recommended or most popular; buyers must self-select without a nudge toward Team.
  • Annual discount offered — There is no annual billing toggle or discount mentioned anywhere on the page.