Augment Code

Included quota

Trial: 30,000 credits (one-time, credit card required). Business ($100/mo flat): up to 50 seats, $100 of usage included, Context Engine, Remote Agents, CLI automation, MCP + 100+ native tools, Code Review, SOC 2 Type II, no-training-on-your-code. Enterprise (custom): bespoke credit limits, SSO/OIDC/SCIM, volume-based annual discounts, Slack integration, ISO 42001 + CMEK compliance, security reports, dedicated support. (The former Indie $20 / Standard $60 / Max $200 individual credit tiers were retired in 2026.)

What is public

Augment Code (augmentcode.com - AI coding platform for large/complex codebases, built on its Context Engine that indexes 100K-500K files across dozens of repos) consolidated its pricing in 2026: the official page now shows ONE flat Business plan plus custom Enterprise. As of June 2026, Business is $100/mo flat (up to 50 seats, with $100 of usage included); Enterprise is custom (bespoke credit limits, SSO/SCIM, CMEK, ISO 42001). The official page frames the $100 in dollars: LLM at the provider's public API list price + a flat 40% service fee on LLM usage (no fee on compute) + Cosmos compute time; credit-based underneath (since Oct 20, 2025) - different models/operations consume different credits. A 30,000-credit trial (credit card required) is the entry; there is no permanent free monthly tier. (The former Indie $20 / Standard $60 / Max $200 individual tiers were RETIRED in 2026 after a March 31, 2026 completions sunset; some third-party listings still show that old 5-tier model.)

Billing mechanics

Credit-based metering (since Oct 20, 2025): a 'credit' reflects actual AI resource consumption, so cost varies by model and task complexity (e.g., a 'fix a 500 error' task runs ~293 credits on Sonnet, ~88 on Haiku, ~488 on Opus; small-to-complex tasks span ~14.5x, and tool calls drive most consumption). The current Business plan is a flat $100/mo (up to 50 seats) with $100 of usage included - the official page meters that $100 in dollars as LLM list price + a flat 40% service fee on LLM (no fee on compute) + Cosmos compute. Beyond the included $100, auto top-up bills $15 per 24,000 credits (~$0.000625/credit, pay-as-you-go at plan rates). Plan credits don't roll over; purchased top-up credits stay valid 12 months. Credits/usage are pooled across the team (no per-seat charge).

Variable cost rationale

High and model-dependent - credits scale with task complexity and (heavily) with model choice; everyday completions are light, but daily Agent use runs ~$60-$200/mo of usage and Remote-Agent/CLI power users exceed $200+/mo; the flat Business base ($100 incl. usage) caps the floor but auto top-up scales the ceiling

Additional watchouts

Repeated, trust-eroding pricing churn in 2025-26 (message-to-credit, completions sunset, individual-tier retirement - 'following Cursor's footsteps' per users); credit-burn anxiety for moderate users; you must actively manage model selection to control cost; only justified for large multi-repo/legacy codebases (overkill + overpriced if your context fits in a smaller window); no permanent free monthly tier

Overage / add-ons

When the included usage/credit pool is exhausted, auto top-up kicks in at $15 per 24,000 credits (same pay-as-you-go rate as your plan, no throttling/slow mode); top-up credits are valid 12 months; monthly plan credits reset with no rollover. Enterprise negotiates bespoke credit limits + volume discounts.

Sales call required

Mixed (some tiers require a call)

Free / trial

30,000-credit trial (card required)

Lowest paid plan

Business $100/mo flat (up to 50 seats)

Commercial notes

Differentiator is the Context Engine - indexes 100K-500K files across dozens of repos (unmatched scale in the AI-coding space), built for large/legacy monorepos; also ships Remote Agents, CLI automation, and Intent (agent-first, multi-agent Coordinator/Implementor/Verifier, macOS); median buyer ~$1,000/yr (community-reported); competes with Cursor, GitHub Copilot (6x cheaper at entry), Claude Code Max (the flat-rate alternative departing users cite), Tabnine, Devin

Key ambiguities

Pricing changed repeatedly and recently - message-based to credit-based (Oct 2025), inline completions sunset on individual tiers (March 31, 2026), then the Indie/Standard/Max individual plans RETIRED, leaving Business ($100/mo flat) + Enterprise (per the official page, June 2026); some third-party listings still show the old 5-tier credit model (Trial/Indie $20/Standard $60/Max $200/Enterprise); per-task credit consumption isn't published as benchmarks (request from sales)

Cancellation / refund

30,000-credit trial (credit card required); Business is self-serve flat $100/mo; Enterprise is custom-contracted (annual, volume discounts); top-up credits valid 12 months; plan credits don't roll over

Support SLA / resale

Standard support (Business); Enterprise adds dedicated support, SSO/OIDC/SCIM, Slack integration, security reports, ISO 42001 + CMEK; SOC 2 Type II and no-training-on-your-code across paid plans; integrates with VS Code + JetBrains, GitHub for code review; MCP + 100+ native tools

Missing data

Per-task/model credit-consumption benchmarks aren't published (request from sales). As of June 2026 the individual Indie/Standard/Max tiers were retired (following a March 31, 2026 completions sunset); the official page now shows a single flat Business plan ($100/mo, up to 50 seats, $100 usage included) plus custom Enterprise, with no permanent free monthly tier (30,000-credit trial only).