Vendor

OpenAI, and why this technical record is deliberately incomplete

OpenAI builds ChatGPT, and its developer documentation layer registers three GPT-5.6 models with pricing and context-window specifications: Sol at $5 and $30, Terra at $2 and $12, Luna at $0.20 and $1.20. No figure on ChatGPT itself can be extracted from this record, and the engineering reason is documented further down.

  • three GPT-5.6 models
  • 1.05M-token operating window
  • consumer surface unmeasured

Last checked: This is a reference page. It is re-checked against the vendor sources and updated when a new version ships. Change log

The state of this maker's versions

Versions recorded here19
  1. 12
  2. 7
  • legacy 12
  • current 7

These are the versions recorded in this reference, not the maker's whole catalogue.

Products

ChatGPT and Sora are logged in our dataset with no independent record, because no fact from the vendor own primary source could be extracted for either.

Model families

Versions

Versions Deployed Status Model families
GPT-4o mini TTS current
GPT-4o mini legacy GPT-4
GPT-4o legacy GPT-4
GPT-5.2 legacy GPT-5
GPT-5.4 legacy GPT-5
GPT-5.5 legacy GPT-5
GPT-5.6 Luna current GPT-5.6
GPT-5.6 Sol current GPT-5.6
GPT-5.6 Terra current GPT-5.6
GPT-5 mini legacy GPT-5
GPT-5 nano legacy GPT-5
GPT-5 legacy GPT-5
GPT Image 2 current GPT Image
o1-pro legacy OpenAI o-series
o1 legacy OpenAI o-series
o3-pro legacy OpenAI o-series
o3 legacy OpenAI o-series
Sora 2 Pro current Sora
Sora 2 current Sora

The segment actually extracted from the primary source

The API model specification registers three frontier units, each carrying an identifier, a price, and a knowledge-cutoff date. GPT-5.6 Sol is specified for complex professional workloads at $5 input and $30 output. Terra sits at the intelligence-to-cost balance point at $2 and $12. Luna is configured for cost-sensitive workloads at $0.20 and $1.20. All three units carry a 1.05M-token operating window, an output ceiling of up to 128K tokens, and a knowledge-cutoff date of 16 February 2026.

Alongside these three, the specification registers GPT Image 2 for image generation, the realtime model line for live voice processing, transcription units, and a text-to-speech unit. The developer platform runs three subsystems: the API itself, Codex as a coding agent operating across desktop, web, CLI, and cloud, and plugins implemented as skills and MCP servers for ChatGPT and Codex.

The segment that stayed out of reach, and the engineering reason why

Every path on openai.com returns a 403 status to this server. We tested this with a browser user agent as well, and repeated it again on 12 August 2026: the domain root and the ChatGPT pricing record both returned 403, while robots.txt on that same domain opens with a 200 status and does not disallow crawling. The obstacle is a bot-protection layer, not a rule we would be circumventing.

The consequence is that ChatGPT plan pricing specifications, its feature set, its geographic-coverage specification, and even the current version identifier for Sora are unavailable from the vendor own primary source. We could have written them from memory and the record would look complete. That is the method the other round-ups run, and it is precisely what this reference was engineered to replace.

The condition under which this gap closes

The day openai.com returns a 200 status to this server. On that day this page, the ChatGPT record, and the Iran column all get updated with a fresh read date, and the change is logged. Until then, the empty cell is our reliable output.

Documented strengths

  • API documentation is transparent: identifier, price, context window, and knowledge cutoff for all three models on one record
  • Three pricing tiers within one model generation, from $0.20 to $5 of input, which simplifies the cost decision
  • A 1.05M-token window on all three models, larger than the one-million-token capacity of its direct rivals
  • Codex, an official coding agent operating across desktop, web, CLI, and cloud infrastructure

Known weaknesses and limits

  • No fact about ChatGPT appears on this record, because the consumer surface at openai.com returns 403 to this server and we do not record from memory.
  • The rates registered here are API rates, not what a ChatGPT subscriber pays. The two layers should not be conflated.
  • The Sora model identifier is absent from the API specification record, so we hold no measurable data on OpenAI video output.
  • The geographic-coverage specification has not been read, so the Iran access column for this vendor stays unchecked.
  • For that reason, the ChatGPT unit sits in our rankings with every cell empty and earns no rank.

Technical verdict

For a developer operating against the API layer, OpenAI is fully assessable and its documentation is built cleanly. For someone evaluating ChatGPT as a purchase, this record is not useful today, and we would rather state that limitation than publish a figure with no source behind it.

Frequently raised questions, with documented answers

Why is the ChatGPT subscription price missing from this record

Because its pricing record could not be read by this server. openai.com returns 403 even with a browser header, while robots.txt does not disallow crawling. A figure not extracted from the vendor own primary source does not get published in this reference.

So is GPT-5.6 the same thing as ChatGPT

No. GPT-5.6 is the identifier for a model line with three versions, Sol, Terra, and Luna, distributed through the API. ChatGPT is a separate product, and which model runs its infrastructure today is exactly the data point that would need to be extracted from the product own record, which is the same record that will not open for this server.

Sources

  1. OpenAI API modelsvendor sourcedevelopers.openai.comread on 12 August 2026
  2. OpenAI developer platformvendor sourcedevelopers.openai.comread on 12 August 2026
  3. openai.com robots.txtvendor sourceopenai.comread on 12 August 2026

How much this tool is searched for

"openai" is searched 2,240,000 times a month in the United States and 49,500 times in Saudi Arabia. That is a 45x gap between two markets this site publishes in.

MarketTerm enteredMonthly search volumeClick costRanking difficultyUser goal
Turkeyopenai165,000$0.26100Navigational
Saudi Arabiaopenai49,500$0.5589Navigational
United Statesopenai2,240,000$0.33100Navigational

Data from Semrush, TR, SA, US databases, retrieved 27 August 2026. Search volumes decay over time; read anything more than three months old as a trend line, not a fixed figure.