AI Stack DB

ChatGPT Review: What It Does, Pricing, and Alternatives

Draft v0.1 — 2026-05-27 KST. content_status = qa_passed. Generated from templates/tool-page-template.md and walked through Section A of qa/adsense-seo-quality-gate.md. Meta description (≤ 155 chars): ChatGPT is OpenAI's general-purpose conversational AI for writing, coding, and everyday questions — what it does, its caveats, and how it compares.

Quick verdict

ChatGPT Health decision check (2026-07-26).

OpenAI announced "Health in ChatGPT" on 2026-07-23, per the official OpenAI RSS feed and the official product article on openai.com. If you already use ordinary ChatGPT for drafting, explaining, or coding help, do not assume Health is just "ask ChatGPT a health question" — per OpenAI's own product article, Health is described as a separate space layered on top of the general assistant reviewed on this page, and it deserves its own decision before you connect anything to it.

Separate surface, not ordinary ChatGPT. OpenAI's product article states Health conversations happen in a distinct space and are not used to train OpenAI's foundation models, and that Health information and memories do not flow back into your non-Health chats — while non-Health context may be used inside Health. Treat Health as a distinct product surface with its own data path, not an extension of the general-purpose ChatGPT pricing, plans, and use cases described earlier on this page.

Supported role, explicit clinical-review boundary. Per OpenAI's own framing, Health is designed to support, not replace, medical care, and is not intended for diagnosis or treatment. That is OpenAI's own product-safety positioning, stated here as a vendor claim rather than as medical guidance from this page. Nothing on this page should be read as advice about whether Health is appropriate for any specific health situation.

Data and integration checklist, grounded in the official privacy notice. OpenAI's Health Privacy Notice (openai.com/policies/health-privacy-policy/) contains medical-record and Apple Health-specific references. Before connecting anything, read that notice directly and check, at minimum: what categories of health data each integration (medical records, Apple Health) can read; whether connected data is used to personalize Health specifically, versus shared elsewhere in the product; how to view and delete Health-specific memories, which OpenAI's product article states users can do; and how to disconnect a health integration later, which the article also describes as available. This page draws only the checklist items from the notice — it does not restate retention periods, encryption specifics, or legal conclusions, because those were not independently re-confirmed here.

Region, device, plan, and availability must be verified live. Per OpenAI's product article, medical-record integrations and some Health apps are U.S.-only, and Apple Health integration requires iOS. Availability by country, device, and ChatGPT plan tier changes over time and is exactly the kind of volatile detail this page routes to the official site rather than asserting as fixed — confirm your own eligibility on OpenAI's official pages before assuming Health applies to your account.

Six-point pre-connection trial checklist. Before linking medical records or Apple Health to Health in ChatGPT, verify each of the following directly on OpenAI's official Health pages and your own account settings:

1. Confirm your country and account are currently eligible for the medical-record and Apple Health integrations you want to use. 2. Confirm your device meets the stated requirement (iOS for Apple Health) before attempting to connect it. 3. Read the current Health Privacy Notice in full for the specific integration you are about to enable, not a summary of it. 4. Identify exactly which data categories (e.g., specific medical-record fields, specific Apple Health metrics) the integration will read, and decide whether you are comfortable sharing each one. 5. Locate the controls to view and delete Health memories, and the control to disconnect the integration, before you connect anything — so you know how to reverse the decision. 6. Decide in advance who reviews any Health-derived output before you act on it, exactly as you would treat any other ChatGPT draft under this page's "Cons and caveats" and "Buyer control and the review boundary" sections below.

Emergency, diagnosis, and treatment boundary. This is product-safety framing, not medical advice: Health in ChatGPT is not an emergency service and is not positioned by OpenAI as a diagnostic or treatment tool. If you are facing a medical emergency, contact emergency services or a qualified clinician directly — do not substitute a chat product for that call. For any non-emergency health decision, OpenAI's own positioning is that Health is meant to support a conversation with your care team, not to replace it.

A launch announcement is not proof of fitness. OpenAI publishing a rollout post about Health in ChatGPT confirms that the feature exists and describes how OpenAI frames it — it does not, by itself, establish clinical accuracy, real-world health outcomes, HIPAA coverage for your specific situation, regulatory compliance in your jurisdiction, endorsement by any healthcare provider, or suitability for any particular medical condition. Each of those is a separate question you would need to verify independently, likely with your own clinician or compliance advisor, before relying on Health for anything beyond a starting point for conversation.

Source-backed evidence note (2026-07-26 KST). Durable facts above come from OpenAI's official RSS feed (openai.com/news/rss.xml, HTTP 200) and the official product article "Introducing ChatGPT Health" (openai.com/index/introducing-chatgpt-health/, HTTP 200, full body read), plus the official Health Privacy Notice (openai.com/policies/health-privacy-policy/, HTTP 200). The RSS-linked article URL (openai.com/index/health-in-chatgpt) returned a Cloudflare challenge to this environment and its body is not treated as read — only the RSS title, publish date, and description are relied on for that specific link. No price, plan entitlement, quota, benchmark, ranking, superiority, encryption guarantee, legal-compliance conclusion, retention period, or availability detail beyond what these three official sources state is asserted here; all of it is vendor evidence, not an independent clinical or regulatory assessment.

GPT-5.6 API-versus-ChatGPT decision check (2026-08-03).

OpenAI published an official announcement on 2026-07-30, per the official OpenAI RSS feed, titled "Advancing the price-performance frontier with GPT-5.6," describing lower GPT-5.6 API pricing. If you use ChatGPT as reviewed on this page, do not assume this API pricing news changes your ChatGPT subscription price, plan, or entitlements — the two are separate products with separate billing, and one cannot be inferred from the other.

Two different products, two different bills. ChatGPT (the consumer/business chat product reviewed on this page, priced under the Free/Go/Plus/Pro/Business/Enterprise plan names above) is billed as a flat-rate or per-seat subscription. The OpenAI API — the developer platform where you call models like GPT-5.6 Sol programmatically — is billed based on tokens used (with additional metrics or per-tool-call fees for some model/tool types), based on how much text you send and receive. A change in per-token API pricing, or the release of a new API model, does not by itself change what a ChatGPT subscriber pays or which models a ChatGPT plan can reach. If you are deciding whether to keep, upgrade, or downgrade your ChatGPT subscription, API pricing news is not evidence either way — check chatgpt.com/pricing/ directly for current plan pricing and entitlements.

What is confirmed about GPT-5.6 API pricing. Per the official OpenAI developer docs (developers.openai.com/api/docs/models/gpt-5.6-sol, read directly on 2026-08-03), GPT-5.6 Sol is described as the frontier model in the GPT-5.6 family, and the gpt-5.6 alias routes API requests to GPT-5.6 Sol. The model page states a context window of 1,050,000 tokens, a maximum input of 922,000 tokens, and a maximum output of 128,000 tokens. Per the official pricing page (developers.openai.com/api/docs/pricing, read directly on 2026-08-03), standard short-context API rates for GPT-5.6 Sol are $5.00 per 1M input tokens, $0.50 per 1M cached input tokens, and $30.00 per 1M output tokens; the official pricing page states that prompts with more than 272K input tokens are priced at 2x input and 1.5x output for the full request, yielding a documented long-context tier of $10.00 / $1.00 / $45.00 per 1M tokens for input / cached input / output, and Batch and Fast Mode rates differ again. These are API developer-platform prices, not ChatGPT subscription prices — verify these rates and the 272K-token threshold, and any rate change since 2026-08-03, directly on the official pricing page before budgeting.

Why the RSS-linked article is not used as evidence here. The RSS item's own description states the announcement covers lower GPT-5.6 pricing for "Luna and Terra" — internal or partner-facing names OpenAI did not define in the RSS text itself, and this page does not guess what they refer to. The linked launch article (openai.com/index/advancing-the-price-performance-frontier-with-gpt-5-6) returned a Cloudflare challenge to this environment, so only the RSS title, publish date, and description are relied on for that item — its body is not treated as read, and no claim here rests on unread article text.

A model-alias or price change is a developer decision, not a ChatGPT-subscriber decision. If you build on the API, a new model version or price point can change your per-request cost and output behavior even when your code sends the same alias (gpt-5.6 today may resolve to a different underlying model tomorrow). If you use ChatGPT as an end user through the web or app, none of this changes your monthly subscription price or which ChatGPT plan you should buy — that decision is answered by chatgpt.com/pricing/, not by API model pages.

Six-step pre-production API evaluation checklist. Before switching a workload to a new model alias or version on the OpenAI API, verify each of the following directly against your own workload and the official docs, not against this summary:

1. Characterize your actual workload shape — typical prompt length, how much of each request is repeated/cacheable context versus fresh input, and typical response length — since price is driven by token counts in each category, not by a single "per request" figure. 2. Confirm current per-token rates for input, cached input, and output directly on the official pricing page before estimating cost, since rates can change between checks. 3. Check whether any of your requests will cross into the long-context pricing tier, and read the official docs for the exact input-length threshold that applies, rather than assuming this page's example rates cover your case. 4. Confirm the model's maximum output token cap fits your use case (128,000 max output tokens is documented for GPT-5.6 Sol as of this check) — a workload that needs longer single responses may need a different model or a chunking strategy. 5. Set a hard budget limit and usage/cost observability (dashboard alerts, per-key spend caps) before routing production traffic to a new model or alias, so an unexpected token-volume spike does not produce an unexpected bill. 6. Run a human evaluation pass on real output samples from the new model/alias against your quality bar before fully cutting over — a lower price per token is not evidence of equal or better output quality for your specific task.

Source-backed evidence note (2026-08-03 KST). This section's API pricing and model-limit claims come from two official OpenAI developer-docs pages read directly on 2026-08-03: developers.openai.com/api/docs/models/gpt-5.6-sol and developers.openai.com/api/docs/pricing, both returning real readable content (not a Cloudflare shell). The announcement date and headline framing come from the official OpenAI RSS feed (openai.com/news/rss.xml), item dated 2026-07-30. The RSS-linked launch article returned a Cloudflare challenge and its body is not treated as read. No ChatGPT subscription price, plan entitlement, quota, benchmark, ranking, speed, accuracy, security, legal, availability, or superiority claim is asserted from any of this; API pricing and ChatGPT subscription pricing/entitlements are separate facts and must not be inferred from one another. All figures above are vendor-published API pricing, stated as vendor evidence, current only as of the dates given.

Source-freshness note (2026-06-24)

If you are sizing up ChatGPT as a general-purpose assistant workflow — drafting, explaining, summarizing, and talking through code in one familiar chat box — decide it against OpenAI's own surfaces rather than any figure quoted on this page.

Source-backed freshness marker. On the 2026-06-22 KST official-source recheck (recorded in data/source-refetch-2026-06-22-evening-gates.json), an automated re-fetch confirmed only that the official ChatGPT homepage https://chatgpt.com/ was reachable (HTTP 200, page title "ChatGPT"); the pricing page https://chatgpt.com/pricing/ and https://openai.com/news/ both returned HTTP 403 (bot-protection shells) in the same pass and are not used as evidence. So the only new durable fact this tick is homepage reachability/title — every price, quota, plan feature, model-availability, and security/privacy detail stays routed to OpenAI's official pages, and the durable plan-structure facts on this page continue to rest on the earlier src-openai-chatgpt-pricing-2026-06-02 page-body read and the 2026-06-15 reachability recheck. This is vendor-surface evidence only — not an independent ranking or benchmark — and no benchmark, ranking, speed, accuracy, or superiority claim is asserted from it.

Pricing-source freshness note (2026-06-15)

On the 2026-06-15 official-source recheck, the official ChatGPT pricing page was reachable again from both entry points. The durable, non-volatile facts confirmed this tick are:

What this recheck does not establish: current USD prices, free-tier quotas, message/usage limits, which models each plan reaches, or per-plan feature entitlements. No amount, quota, or model-availability detail is asserted from it — all such specifics remain "verify on the official site." This is a source-backed freshness marker confirming the official pricing page is reachable again; it adds no new pricing claim over the 2026-06-02 page-body read below. (https://openai.com/news/ returned HTTP 403 in the same pass and is not used as evidence here.)

Source-freshness note (2026-06-19)

If you are weighing ChatGPT as a general-purpose assistant workflow — drafting, explaining, summarizing, and talking through code in one familiar chat box — the durable way to decide is to check OpenAI's own surfaces yourself rather than trust any figure quoted on this page.

Source-backed freshness marker. On the 2026-06-19 KST recheck, an automated re-fetch of the OpenAI/ChatGPT surfaces from this environment returned HTTP 403 (bot protection), so no new live reachability claim is made today — the durable plan-structure facts on this page continue to rest on the earlier official-source reads (the 2026-06-02 chatgpt.com/pricing/ page-body read and the 2026-06-08 / 2026-06-15 reachability rechecks), and every volatile specific stays routed to the official site. This is vendor-surface guidance reported as vendor evidence only — not an independent ranking or benchmark — and no USD price, quota, message limit, per-plan feature, or model-availability claim is asserted from it.

Pricing-source freshness note (2026-06-08)

On the 2026-06-08 official-source recheck, the ChatGPT pricing page was reachable from both entry points. The durable, non-volatile facts confirmed this tick are:

What this recheck does not establish: current USD prices, free-tier quotas, message/usage limits, which models each plan reaches, or per-plan feature entitlements. No amount, quota, or model-availability detail is asserted from it — all such specifics remain "verify on the official site." This is a source-backed freshness marker confirming the official pricing page is reachable again; it adds no new pricing claim over the 2026-06-02 read below.

Latest pricing-source check (2026-06-02)

The official ChatGPT pricing page at https://chatgpt.com/pricing/ — which returned HTTP 403 to automated requests on the 2026-05-27 check — was reachable (HTTP 200) on 2026-06-02. From that page-body read, the durable, non-volatile facts are:

What this check does not establish: specific prices, free-tier quotas, message/usage limits, which models each plan reaches, or per-plan feature entitlements. The saved read contained no quotable USD price figures, so none are stated here — all amounts and limits remain "verify on the official site."

This is a source-backed freshness note. The pricing summary on this page has moved from "blocked (403)" to "official pricing page reachable," but no price or quota is asserted. Confirm current plans, prices, limits, and model availability on the official site before relying on them.

What is ChatGPT?

ChatGPT is the general-purpose conversational AI product from OpenAI. You type a question or instruction in plain language and it replies in plain language, holding the thread across a back-and-forth conversation. It is the product most people picture when they hear "AI chatbot," and it lives at chatgpt.com with companion mobile and desktop apps. OpenAI also exposes its underlying models through a separate developer API, which is a different surface from the consumer chat app described here.

The practical pitch is breadth rather than a single specialty: one chat box you can aim at writing, editing, explaining, coding help, summarizing, and everyday questions. That generality is the whole point — and also the reason it is worth pairing with more specialized tools for jobs where a focused product fits better (cited research, in-IDE coding, in-suite productivity).

The exact model lineup, which models a given plan reaches, message limits, available connectors/apps, and the enterprise/data-handling terms all shift between releases. This page deliberately does not pin those down, because they are volatile. The official pricing page was reachable on 2026-06-02 and confirms the plan tier names and the free-vs-paid billing structure (see the freshness note above), but it surfaced no quotable prices or per-plan feature detail. Treat anything specific — a model name, a price, a quota — as something to confirm on OpenAI's own pages before relying on it.

Main use cases

Pricing and plans

OpenAI offers ChatGPT as a freemium consumer product (free access plus paid subscription tiers) and, separately, a usage-based developer API. That much is long-standing and stable positioning.

The official pricing page at https://chatgpt.com/pricing/ was reachable on 2026-06-02 and again on the 2026-06-08 and 2026-06-15 rechecks (HTTP 200, from both entry points) — an improvement over the 2026-05-27 check, when both chatgpt.com/pricing/ and openai.com/chatgpt/pricing/ returned HTTP 403. From the 2026-06-02 read, the plan structure visible on the page is:

Beyond that structure, no monthly price, message/usage limit, model-access detail, or connector/app entitlement is quoted on this page. The saved read contained no quotable USD price figures, and prices, quotas, and model availability all change frequently.

Plan names, prices, free-tier limits, and per-plan features should be verified on the official website at https://chatgpt.com/pricing/ before you rely on them. Model availability and included features within each plan change frequently.

The developer API is priced separately from the consumer subscription and on a different page; nothing about API token rates is asserted here either.

Pros

Cons and caveats

Alternatives

Buyer control and the review boundary

If you are evaluating ChatGPT as a buyer rather than a casual user, the deciding question is less "can it write and explain?" than "which of your content does ChatGPT receive or retain, and where is the review boundary before its output is published or acted on?" <span id="chatgpt-review-boundary-2026-06-28"></span>

Evergreen decision framing only. No price, plan, quota, model-availability, benchmark, ranking, speed, accuracy, superiority, or legal claim is made here; verify current plan inclusions, data-handling, and security/enterprise terms on OpenAI's official pages.

Where to compare ChatGPT next

If you have already decided ChatGPT is a contender, the next question is usually "against what, and for which job?" These side-by-side pages are organized by workflow fit, not by a winner — each one walks through where one tool's shape suits a particular task better than the other, so you can follow the path that matches your own work:

To browse the whole field rather than a single head-to-head, start from the AI Assistants category. These links are decision paths, not rankings — no benchmark, price, quota, or model-availability claim is made here; the comparison pages route any such specifics to the official sources.

Who should not use ChatGPT

Author selection rubric

Choose ChatGPT when at least two of these are true:

Avoid ChatGPT (or pair it with something else) when any of these are true:

Sources

Sources flagged needs_verification or blocked are not quoted as fact above; everything specific (prices, limits, model availability) is routed to "verify on the official site." The 2026-06-02 ok pricing source is relied on only for the plan tier names and billing structure, not for any amount.

Internal links (at least 3)

Disclosure

Trademark notice

ChatGPT and OpenAI are trademarks of OpenAI. Use here is referential only and does not imply endorsement, partnership, or affiliation.

Update log