Team discussing SaaS subscription plans with laptops in a modern office

In 2026 OpenAI expanded personal tiers to Free → Go → Plus → Pro, with Business and Enterprise on the team side. When people search “ChatGPT Work pricing,” the real decision blockers are usually not the monthly fee alone: what Work includes, how it shares Codex quota, whether free is enough, and when Enterprise is mandatory.

This guide lists official public pricing (US reference; other regions are localized) and includes a dedicated ChatGPT Work access comparison table. By the end you can decide whether to stay on free, move to Plus/Pro, or negotiate Business/Enterprise — with FAQ and anti-patterns at the close.

Reading path: §1 what Work is → §2 six-tier price table → §3 Work/Codex quotas → §4–§5 individual vs team → §6 decision table → §7 when not to upgrade.

1. What is ChatGPT Work? How it relates to Plus and Codex

ChatGPT Work is OpenAI’s productivity feature set inside the ChatGPT client for office and knowledge work — web tasks, deep desktop integration, and a shared usage pool with Codex (the coding agent). Per OpenAI’s official docs, Work and Codex quotas, billing, and caps are linked. Upgrading mainly buys higher model access + larger Work/Codex usage, not a separate “Work plugin.”

Unlike the old “ChatGPT = chat only” era, the 2026 lineup is clearest in three layers:

  • Core chat: all tiers, but free has hard caps on messages, uploads, image generation, and deep research.
  • ChatGPT Work: projects, scheduled tasks, custom GPTs, Canvas, data analysis — “get work done” features. Free gets limited desktop-only access; Plus expands to web and mobile.
  • Codex: IDE extension, CLI, cloud Agent. Pro is labeled “Maximum Codex tasks”; Business/Enterprise can buy extra credits.

When discussing “Work pricing,” always include Codex: heavy developers often hit Codex limits before chat message caps.

2. 2026 subscription pricing overview (including Go and Business)

Prices below are US reference (July 2026); India, Korea, EU, etc. adjust for local currency and tax — checkout page is final. Enterprise has no public list price.

TierMonthly (US ref.)BillingMin seatsBest for
Free$01Trial, low-frequency Q&A
Go$8Monthly1More messages than free, not full Plus
Plus$20Monthly1Daily office work, light dev, cross-device Work
Pro$200Monthly1Researchers, indie devs, heavy Codex
Business$25 (annual) / $30 (monthly)Per user2Small teams, shared workspace, SSO
EnterpriseCustom (often $40–$60/user/mo est.)Annual contractNegotiatedCompliance, data residency, large seats

Source: openai.com/chatgpt/pricing and OpenAI’s Go launch; Enterprise range is common third-party estimate, not public list price.

2.1 The easy-to-miss “Go” tier

ChatGPT Go ($8/month), rolled out globally in early 2026, fills the gap between Free and Plus: significantly more messages, uploads, and images than free, but without Plus-level reasoning models and full cross-device Work. If you only need a few hundred extra chats per month and no complex projects, Go beats jumping straight to Plus; once you need unified Work + custom GPTs across desktop/web/mobile, Plus is the better fit.

3. Work access and Codex quota by tier

This is the core of plan selection: each monthly fee buys a Work entry point plus a usage ceiling.

TierChatGPT WorkCodexTypical bottleneck
FreeDesktop only, limitedLimitedMessage caps, no mobile Work
GoNot full cross-device Work (see official compare)Higher than FreeNo Plus-level reasoning/projects
PlusDesktop + web + mobile expandedExpanded usageLong-running Codex Agent jobs
ProPlus capabilities + higher total usageMaximum task tierPrice ($200/month)
BusinessTeam workspace + shared GPTsPer seat; can buy credits32K non-reasoning context (large repos)
EnterpriseAbove + admin consoleFlexible credit-based scalingSales cycle, minimum commitment

When Business/Enterprise Work or Codex usage spikes, you can buy workspace credits without upgrading the whole team — a variable finance should align on early.

4. Free vs Pro: how individual users should choose

4.1 When free is enough

  • Fewer than ~20 serious conversations per week; no long-document uploads or data analysis.
  • Occasional Work on a desktop only, accepting queues and rate limits.
  • Coding is mostly copy-paste snippets — no long IDE Codex agent runs.

4.2 Signals to upgrade to Plus ($20)

  • Same Work flow on phone, browser, and desktop (projects, scheduled tasks, custom GPTs).
  • Weekly PDF/spreadsheet analysis or reliance on GPT-5.6-class reasoning models.
  • Moderate Codex: PR edits, scripts — but not “agent running all day.”

4.3 Signals to consider Pro ($200)

  • ChatGPT as primary dev environment: Codex CLI, cloud Agent, deep research hitting Plus caps daily.
  • Need Pro-only reasoning modes (e.g. GPT-5.6 Sol Pro) and higher image/research quotas.
  • Personal income or project margin covers $200×12 = $2,400/year with measurable time savings.
Anti-pattern: occasional email polish or copy editing on Pro — Plus covers ~90% of cases; the extra $180/month rarely pays back.

5. Business vs Enterprise: team pricing and differences

5.1 ChatGPT Business (formerly Team)

Rebranded mid-2025 as Business. Core value: shared workspace, admin console, SAML SSO, SOC 2 Type II, and conversations not used for model training by default. Pricing $25/user/month (annual) or $30/user/month (monthly), minimum 2 seats — a two-person team on annual billing starts around $600/year.

vs “everyone on Plus”: Business costs more mainly for governance and collaboration, not a sudden IQ jump. A 5-person team without SSO might still be cheaper at 5×$20 Plus (but lacks unified billing and shared GPT policy).

5.2 ChatGPT Enterprise

Enterprise has no public price — sales quote required. vs Business it typically adds:

  • ISO 27001 series, SCIM provisioning, RBAC, IP allow lists
  • Multi-region data residency (US, EU, UK, Japan, Korea, Singapore, etc.)
  • 24/7 priority support, custom legal terms, flexible credits billing

Third-party estimates for a 50-person team often land around $30,000–$36,000/year ($50–$60/seat/month) — contract terms win. Ask whether unlimited Codex is included, credit unit price, and data retention period.

5.3 When Enterprise is actually “worth it”

Negotiate Enterprise only if at least one applies; otherwise Business or scattered Plus is often more flexible:

  1. Security requires SCIM + fine-grained GPT permissions + audit logs.
  2. Legal requires data in a specific country/region.
  3. 50+ seats with unified SLA and customer success manager.

6. Pick by persona: one decision table

Who you areSuggested tierMonthly ballpark
Student / occasional researchFree or Go$0–8
PM, ops, light writerPlus$20
Full-time indie dev, AI researcherPro$200
2–30 person startup needing SSOBusiness$50–900+
Finance, healthcare, large R&D orgEnterpriseCustom
Heavy API embedding in productAPI pay-as-you-go (not ChatGPT sub)Token-based

Building SaaS or automation pipelines? ChatGPT subscription and OpenAI API token billing are separate books: subscription for humans clicking the client; API for 24/7 programmatic calls. Mixing them is a common overspend cause.

7. Three common buying mistakes (anti-patterns)

7.1 Two-person startup buys Business “to look legit”

2×$30=$60/month just for SSO when two people underuse collaboration — two Plus accounts ($40) + 1Password may be cheaper. Join Business when you hit the fifth member.

7.2 Treating Pro as unlimited API

Pro still has abuse protection and fair-use limits. Product-embedded AI and batch offline jobs belong on API or self-hosted inference — not ten Pro accounts in rotation.

7.3 Ignoring Go and giving everyone Plus

Admin and support roles that only polish short text often fit Go or Free; reserve Plus for people who truly need Work + Codex. Total cost drops noticeably.


8. When subscription is not enough: API and local compute

When Codex or Work caps out, engineering teams use three paths besides upgrading:

  1. API offload: batch summarization, classification, and structured tasks via API with cache and Batch discounts (see 2026 AI coding tool cost ranking).
  2. Local agents: run Codex CLI, Claude Code, or Ollama on Mac — long debug loops on-device, cloud subscription only for strongest-model slices.
  3. Team credits: Business/Enterprise buy workspace credits from OpenAI — easier budget control than upgrading everyone to Pro.

If you already run Codex CLI or local models on Mac and need a stable remote macOS environment for CI or long-running agents, dedicated Apple Silicon cloud hosts can absorb peak compute — see Macstripe home options.


9. Frequently asked questions

Is ChatGPT Work the same as ChatGPT Plus?

No. Plus is a $20/month personal tier that includes expanded ChatGPT Work access. ChatGPT Work itself is the productivity feature suite (Codex, projects, tasks, etc.) with different quotas per tier.

Can the free tier use ChatGPT Work?

Yes, but only on desktop with tight limits. For stable cross-platform Work and more Codex tasks on web and mobile, you typically need Plus or higher.

How do Pro and Enterprise differ?

Pro ($200/month) targets individual power users with higher Work/Codex usage. Enterprise is a custom contract adding SSO, SCIM, data residency, compliance certifications, and purchasable extra workspace credits — suited for organizations of 50+.

What is the minimum seat count for Business?

Officially at least 2 seats. Annual billing is about $25/user/month; monthly is about $30/user/month. Small teams that only need personal accounts often find Plus more economical.

What happens when Work quota runs out?

Wait for the cycle reset, upgrade to a higher tier, or purchase extra workspace credits on Business/Enterprise. Long dev tasks can also shift to local Codex CLI + Mac to reduce subscription pressure.

10. Further reading