ChatGPT
Identity card
The company
- Name
- OpenAI ↗
- Founded
- 2015
- Head office
- San Francisco, California, United States
- Chief executive
- Sam Altman
The product
- Flagship model
- GPT-6 Astra
- Subscription
- €22.99 per month · Plus
- API
- 10 $ / 50 $ input / output, per million tokens
The data
- Law
- United States
Usage
- Web traffic, Sept. 2026
- 56.4 %
Key points
ChatGPT is the most used assistant among the general public, and the most complete: images, voice and agents. OpenAI spends far more than it earns, and advertising arrived on the free plan in Europe in August 2026.
- Varied everyday uses
- Images
- First steps with AI
- Avoiding advertising without paying
- Sensitive data on the free plan
Source: Similarweb, share of visits to generative AI sites, worldwide; records published by Similarweb and repeated by the press.
Yes for an all-purpose assistant, especially if you are starting out or want images. Look elsewhere for demanding writing or code work (Claude), or for data kept in Europe (Vibe).
Strengths
- The assistant of 1.2 billion peopleOpenAI claims more than 1.2 billion weekly users in October 2026; ChatGPT accounts for 56.4% of web traffic to AI sites in September 2026.Source: OpenAI, 7 October 2026 ↗ ; OfficeChai, citing Similarweb, 12 October 2026 ↗
- The strongest on reasoning testsIts best model passes 95% of the ARC-AGI-2 test, the best score of the 10, and ranks 1st in the Epoch AI index.Source: ARC Prize, ARC-AGI-2, 28 October 2026 ↗ ; Epoch AI, ECI index, 4 November 2026 ↗
- A broad free planThe free plan gives unlimited written conversations with GPT-6, subject to measures against abuse; images, files and voice are limited.Source: OpenAI, ChatGPT pricing page, 7 November 2026 ↗
Weak points
- A loss-making companyIn 2025, OpenAI spent 34 billion dollars for 13 billion dollars of revenue, according to financial documents reported by the Financial Times; OpenAI has not published audited accounts and did not comment on them.Source: The Next Web, citing the Financial Times, 16 June 2026 ↗
- Advertising on the free planSince late August 2026, users of the free and Go plans see advertisements in 31 European countries, including France.Source: The Next Web, 20 August 2026 ↗
- Family lawsuits brought together in courtSince 3/2/2026, at least 9 lawsuits from families, several of them over suicide, have been joined in a single proceeding in California.Source: Superior Court of California, coordination order, 3 February 2026 ↗
The company OpenAI
The OpenAI entry →Company history
OpenAI was founded in December 2015 in San Francisco, as a non-profit lab, by Sam Altman, Elon Musk, Greg Brockman, Ilya Sutskever and others. Its mission: to ensure that general AI benefits all of humanity. To pay for compute, it creates a for-profit subsidiary and allies with Microsoft.
The story continues
On 30 November 2022, ChatGPT goes online and launches the generative AI wave. OpenAI then releases a succession of models, from GPT-4 in 2023 to GPT-6 in 2026.
In October 2025, the company becomes a public benefit corporation controlled by its foundation; in March 2026, it raises 122 billion dollars.
Sources: Wikipedia, 7 November 2026 ↗ ; OpenAI, 30 November 2022 ↗ ; Microsoft, 28 October 2025 ↗ ; Built In, 1 April 2026 ↗
Leadership
- Chief executive officer and co-founder
- Sam Altman ↗
- President and co-founder
- Greg Brockman ↗
- Chief financial officer
- Sarah Friar ↗
- Chair of the board of directors
- Bret Taylor ↗
Key figures
- Valuation, March 2026
- 852 billion dollarsBuilt In, 1 April 2026 ↗
- Revenue, 2025
- 13 billion dollarsThe Next Web, citing the Financial Times, 16 June 2026 ↗
- Spending, 2025
- 34 billion dollarsThe Next Web, citing the Financial Times, 16 June 2026 ↗
The ChatGPT plans
- FreeFree0 €0 $
- GoStandard7.99 €
- PlusStandard22.99 €20 $
- ProPremiumprice published in dollars only$100 to $500no euro price
- Business, Standard seatTeam, per seat26 €25 $
OpenAI pricing pages: incl. VAT in France, excl. tax in the United States; up to date as of 9 November 2026.
Content of the 5 plans, prices in France and the United States
Prices in France and the United States. In the United States, prices are shown in dollars and excluding tax; in France, they include 20% VAT. So $20 does not equal €20: the French price is often slightly higher. When only the dollar price appears, the company does not publish a euro price: the amount paid in France depends on the exchange rate and VAT.
VibeMistral Large1.50 $0.50 $
PerplexitySonar2.50 $0.25 $
DeepSeekV4-Pro3.96 $1.32 $
Meta AIMuse Spark 1.14.25 $1.25 $
QwenQwen3.8-Max6 $2 $
GrokGrok 4.76 $2 $
GeminiGemini 4 Argon10 $2 $
ClaudeOpus 5.520 $4 $- ChatGPTGPT-6 Astra50 $10 $33 times the output price of Vibe
Copilot: no API of its own. Flagship model of each company, standard rate, up to date as of 9 November 2026.
OpenAI API prices for 3 models
| Model | Input, $ per million tokens | Output, $ per million tokens |
|---|---|---|
| GPT-6 Astra | 10 | 50 |
| GPT-6.1 Sol | 2 | 10 |
| GPT-6 Luna | 0.10 | 0.50 |
Prices up to date as of 9 November 2026, read on the companies' pages, or from 2 concordant sources; the date of each record is in the page's price file Methodology.
The ChatGPT models
- Company
- OpenAI
- Models offered
- GPT-6 Solby default, paid plans (Plus, Pro, Business, Enterprise); 11th out of 71 on general intelligence (official page ↗)GPT-5.6 Lunaby default, Free, Go; 31st out of 71 on general intelligence (official page ↗)GPT-6 Lunaoptional, Free, Go, Plus, Pro; 33rd out of 71 on general intelligence (official page ↗)GPT-5 Thinking Minioptional, Free, Go (Plus and Pro: extended access) (official page ↗)GPT-6.1 Solpaid plan, Plus, Pro; 3rd out of 71 on general intelligence (official page ↗)GPT-6 Astrapaid plan, Plus, Pro; 2nd out of 71 on general intelligence (official page ↗)GPT-6 Solpaid plan, Plus, Pro; 11th out of 71 on general intelligence (official page ↗)GPT-6 Propaid plan, Pro ($200) (official page ↗)GPT-5.6 Terraoptional, Free and Go (limited, Work and Codex on desktop), Plus, Pro; 16th out of 71 on general intelligence (official page ↗)GPT-6.1 Terrapaid plan, Plus, Pro, Business, Enterprise, Edu (ChatGPT Work and Codex); 5th out of 71 on general intelligence (press ↗)GPT-5.6 Sol, GPT-5.6 Luna, legacy models (Legacy)paid plan, Plus, Pro (official page ↗)GPT-5.6 Sol Propaid plan, Pro (official page ↗)
Links read on 10/11/2026 on the company's pages, press as a fallback. Rank: Epoch AI ECI index (CC BY 4.0). Performance of AI models
Data management
How OpenAI handles conversations
- Training models on conversationsFree, Go, Plus and Pro plans: yes by default. Business, Enterprise, Edu and API: no by default.Source: OpenAI, business privacy, 8 January 2026 ↗
- Opting out of trainingTurn off "Improve the model for everyone" in the settings, under Data controls; this applies to subsequent conversations.Source: OpenAI, help centre, 7 November 2026 ↗
- Retention periodHistory is kept until you delete it; temporary conversations are not used for training and are kept for up to 30 days.Source: OpenAI, help centre, 7 November 2026 ↗
- Applicable lawUS law: OpenAI is a US company, subject in particular to the Cloud Act, even when the servers are in Europe.Source: Guide d'usage de l'IA pour les agents publics de l'État (DINUM), June 2026 ↗
Score on this criterion: 33 out of 100, joint 4th out of 10. Training by default, can be turned off; Italian fine of €15m annulled in 2026 for lack of jurisdiction, without examining the substance. The ethics ranking
Environment
Environment score
40 7th out of 10- Simple query 40 %50
- Advanced queries 20 %33
- Decarbonised electricity 20 %33
- Environmental approach 20 %33
Black line: average of the 10. The environmental ranking
Our view
The free version runs on GPT-6 Luna, a small model, the paid version on Sol, and all models reason. The only estimate available, with low confidence: CLEER gives 0.17 Wh for GPT-5.6 Luna and 1.36 Wh for GPT-5.6 Sol, their predecessors. Heavy uses weigh: 700 million images in one week in April 2025, images opened to free users, agents. OpenAI has published a single figure, 0.34 Wh, without a method, and no emissions report. ChatGPT receives 56.4% of visits to AI sites in September 2026: it is the assistant whose total usage weighs the most.
Per-question figure published by OpenAI
0.34 Whper question, according to OpenAI: statement of June 2025, average question, without a method.
Company figures cannot be compared with each other: scopes and methods differ. The "Simple query" criterion of the environmental ranking combines them with external estimates, for lack of a common basis: the methodology sets out the limits.
- Gemini0.24 Wh
Copilotnothing published- Vibe1.14 gCO2e
- Claudenothing published
- ChatGPT0.34 Wh
- Qwennothing published
- Meta AInothing published
- Perplexitynothing published
- DeepSeeknothing published
- Groknothing published
On the Éclairages blog: What is the electricity consumption of AI? ↗ ; Which AIs pollute the most? ↗
Scores and criteria
The ranking →The Celsius score is the average of 3 scores: performance, ethics and environment. Each combines several criteria. Each row places ChatGPT among the 10; a click on a criterion opens its definition and the justification for ChatGPT's score.
5 public rankings 100 %933rd out of 10
Definition. Synthesis of 5 public performance rankings, where models are ranked by tests or by votes from internet users: each ranking gives 100 to the best of the 10 and 0 to the last.
Justification. LMArena: 22nd (rank among 419 models); Artificial Analysis: 57 (index v5.0.1); Epoch AI: 169.1 (ECI index); compar:IA: 4th (rank among 121 models); ARC-AGI-2: 95 (% solved).
Content produced 30 %67joint 2nd out of 10
Definition. Content the assistant produces on request (sexual, misleading images, minors) and results of independent tests.
Justification. Parental controls (9/2025) and age prediction (1/2026); "adult mode" announced then suspended (3/2026); public figures allowed in images (2025).
Use of data 30 %33joint 4th out of 10
Definition. Use of conversations for training, retention period, storage location, sanctions by the authorities.
Justification. Training by default, can be turned off; Italian fine of €15m annulled in 2026 for lack of jurisdiction, without examining the substance.
Safety 20 %672nd out of 10
Definition. Independent safety evaluations (Future of Life Institute, SaferAI, Phare) and resistance to attacks.
Justification. Grade C at the Future of Life Institute, 34% at SaferAI; o3 sabotaged its shutdown 79 times out of 100 (Palisade, 5/2025); at least 9 lawsuits from families joined in a single proceeding in California (2/2026).
Governance 20 %424th out of 10
Definition. Control of the company and deliberate choices: military uses, shareholding, commitments kept or abandoned.
Justification. Agreement with the Pentagon a few hours after Anthropic's sanction (2/2026); $12.5m from its president to MAGA Inc. (2025).
Simple query 40 %50joint 5th out of 10
Definition. Consumption of a typical text query, on the default model served: figure published by the company, third-party estimate, failing that the size of the model.
Justification. Free: GPT-6 Luna, a small model; paid: Sol; all reason. Low-confidence CLEER estimates for their predecessors GPT-5.6: 0.17 and 1.36 Wh.
Advanced queries 20 %33joint 6th out of 10
Definition. Weight of the resource-hungry uses the product promotes: video and image generation, agents, reasoning enabled by default.
Justification. 700 million images in one week (April 2025), images for free users, agents; the Sora video app closed in March 2026.
Decarbonised electricity 20 %33joint 4th out of 10
Definition. Share of low-carbon electricity in the data centres and efforts to increase it: local grid, on-site gas, renewable or nuclear electricity contracts.
Justification. Mostly in the United States (384 g CO₂e/kWh); gas turbines authorised at the Stargate site in Abilene.
Environmental approach 20 %33joint 6th out of 10
Definition. What the company publishes and commits to: energy and emissions figures, environmental report, targets, model optimisation.
Justification. A single figure, 0.34 Wh, without a method (June 2025); no emissions report or climate target.
Compare
or ChatGPT or ClaudeChatGPT does more things and reaches more people; Claude is better rated on performance and ethics.Celsius score: 62 against 74 or ChatGPT or GeminiThe 2 mainstream giants: ChatGPT more used, Gemini more frugal and better integrated with Google.Celsius score: 62 against 71 or ChatGPT or CopilotCopilot is built on OpenAI models: ChatGPT for the standalone assistant, Copilot for Word and Excel.Celsius score: 62 against 58Other comparisons
ChatGPT or Meta AI+14ChatGPT or Vibe+22ChatGPT or Qwen+23ChatGPT or Perplexity+26ChatGPT or DeepSeek+27ChatGPT or Grok+36Celsius score gap.
Ranked 3rd out of 10 in the assistant ranking, scores as of 11 November 2026. Facts and figures in the entry recorded from 7 to 9/11/2026, each with its source; foundation and head office: Wikipedia, 9 November 2026 ↗.
Continue in the ranking
The full ranking →CompanyOpenAI14models measured, of which 10 with an entryIts assistant, its models, its governance and the electricity of its data centres.The entry →ModelOpenAIGPT-6 Astra2ndout of 71 on general intelligenceThe best ranked on general intelligence of the models ChatGPT offers.The entry →Comparison2 AI assistantsChatGPT or Claude62 against 74Celsius score out of 100ChatGPT does more things and reaches more people; Claude is better rated on performance and ethics.The comparison →The hub newsletter
This month's ranking, news on the impact of AI and one figure from the AI Impact Database. One newsletter a month at most, free of charge. Past issues
