OpenAI Releases GPT-5.5, Advancing Its Super App Roadmap

OpenAI released GPT-5.5 on Thursday, presenting it as “the smartest and most intuitive to use model” and framing the launch as “the next step toward a new way of getting work done on a computer.”

“OpenAI just announced that ChatGPT is getting a model upgrade to GPT-5”

9to5Google

In a briefing, OpenAI co-founder and president Greg Brockman said, “This model is a real step forward towards the kind of computing that we expect in the future — but it is one step, and we expect to see many in the future,” and he added that GPT-5.5 is “a faster, sharper thinker for fewer tokens compared to something like 5.4.”

9to5Google

Multiple outlets tied the release to OpenAI’s “super app” roadmap, with Brockman claiming it brings the company “one step closer to the creation of OpenAI’s ‘super app.’”

The model is positioned as more capable across “agentic coding and knowledge work,” as well as “mathematics and scientific research,” with OpenAI and reporters describing it as useful for “writing and debugging code, researching online, analyzing data, creating documents and spreadsheets, operating software, and moving across tools until a task is finished.”

OpenAI’s own product post says GPT-5.5 “understands what you’re trying to do faster and can carry more of the work itself,” and it emphasizes that users can give the model “a messy, multi-part task” and “trust it to plan, use tools, check its work, navigate through ambiguity, and keep going.”

OpenAI also said the rollout begins Thursday for paid tiers in ChatGPT and Codex, while API access is described as “very soon” but gated by additional safety work, including “different safeguards.”

OpenAI’s GPT-5.5 launch is repeatedly linked to a broader product vision that outlets describe as a “super app” built from multiple OpenAI components.

Firstpost says OpenAI co-founder and president Greg Brockman claimed GPT-5.5 “also brings the company one step closer to the creation of an OpenAI super app,” describing the super app as “a multi-purpose program” still “under development.”

Business Today

TechCrunch similarly reports that Brockman said the model brings OpenAI “one step closer to the creation of OpenAI’s ‘super app,’” and it quotes him describing the advancement “towards more agentic and intuitive computing.”

The “super app” concept is described as combining “ChatGPT, Codex, and an AI browser into one unified service that can aid enterprise customers,” a framing echoed across Firstpost and TechCrunch.

CXOToday adds that GPT-5.5 was “Codenamed Spud internally,” and it says Brockman told media that GPT-5.5 provided “more agentic and intuitive computing,” while also calling the model “a real step forward towards the kind of computing that ‘we expect in the future – but it is one step, and we expect to see many in the future.’”

OpenAI’s own release post emphasizes the mechanics behind that vision, saying GPT-5.5 “understands what you’re trying to do faster” and can “carry more of the work itself,” including “operating software” and “moving across tools until a task is finished.”

OpenAI and multiple outlets highlighted benchmark performance and efficiency as central to GPT-5.5’s appeal, repeatedly contrasting it with GPT-5.4 and competitor models.

“Sam Altman-led OpenAI has released its new generation AI model, the GPT-5”

Business Today

OpenAI’s product post includes a table of results, stating that on Terminal-Bench 2.0 GPT-5.5 achieves “82.7%” while GPT-5.4 is “75.1%,” and it lists OSWorld-Verified at “78.7%” for GPT-5.5 versus “75.0%” for GPT-5.4.

The same OpenAI post says on SWE-Bench Pro GPT-5.5 reaches “58.6%,” and it also provides a comparison for GDPval (wins or ties) of “84.9%” for GPT-5.5 against “83.0%” for GPT-5.4.

TechCrunch reports that OpenAI released data showing GPT-5.5 “consistently scores higher” across benchmarks compared with models from “Google and Anthropic,” and it names competitors “Gemini 3.1 Pro and Claude Opus 4.5” in its discussion of comparisons.

The Verge describes GPT-5.5 as “more efficient and better at coding,” and it quotes OpenAI’s guidance that GPT-5.5 can use “significantly fewer” tokens to complete tasks in Codex.

OpenAI’s own post also claims that GPT-5.5 “matches GPT‑5.4 per-token latency in real-world serving” while performing at “a much higher level of intelligence,” and it says it “uses significantly fewer tokens to complete the same Codex tasks.”

Alongside performance claims, outlets emphasized safety guardrails and the conditions for API availability, with OpenAI describing “strongest set of safeguards to date” and multiple reports tying the delay to cybersecurity work.

OpenAI’s product post says it is releasing GPT-5.5 “with our strongest set of safeguards to date,” and it describes evaluating the model “across our full suite of safety and preparedness frameworks,” working with “internal and external redteamers,” and collecting feedback from “nearly 200 trusted early-access partners before release.”

CXOToday

CNBC reports that OpenAI said GPT-5.5 “does not cross its ‘Critical’ cybersecurity risk threshold,” but it “does meet the criteria for its ‘High’ risk classification,” and it quotes Mia Glaese saying GPT-5.5 “underwent extensive third-party safeguard testing and red teaming for cyber and bio [risks].”

Firstpost and TechCrunch both describe that GPT-5.5 includes “stricter classifiers for potential cyber risks,” and Firstpost also says OpenAI deployed the model on “Nvidia GB200 and GB300 NVL72 systems” with “load balancing and partitioning heuristics.”

Axios and CNBC both connect the API timing to additional safety guardrails, with Axios saying API access is coming “once OpenAI finishes incorporating additional cybersecurity guardrails,” and CNBC stating that the company said the model will come to its API “very soon,” but those deployments require “different safeguards.”

OpenAI’s own post confirms the rollout scope, saying “Today, GPT‑5.5 is rolling out to Plus, Pro, Business, and Enterprise users in ChatGPT and Codex,” and that “GPT‑5.5 Pro is rolling out to Pro, Business, and Enterprise users in ChatGPT.”

The release of GPT-5.5 is presented across outlets as part of an accelerating model cadence and a continuing push into enterprise and computer-based work, with multiple reports pointing to the next steps in deployment and broader adoption.

“As part of the ongoing battle for supremacy over the AI ecosystem, OpenAI has released its latest flagship model – GPT 5”

CXOToday

TechCrunch notes that OpenAI released its last model only “last month,” with previous releases in “December and, before that, November,” and it says company staff expected the pace to “continue for the foreseeable future.”

SQ Magazine

It also quotes Jakub Pachocki saying, “We see pretty significant improvements in the short term, extremely significant improvements in the medium term,” and he adds, “I would say, like, I think the last two years have been surprisingly slow.”

The New Stack similarly describes GPT-5.5 as part of a larger shift where “the model itself is no longer the whole product,” and it quotes Brockman saying, “You can think of it as the brain, but also building the body in terms of the applications we ship, the agentic harnesses.”

Axios frames the economic and enterprise implications through Nvidia’s involvement, saying Nvidia says its new chips “slash the cost of running advanced AI like GPT-5.5 up to 35x per token,” and it quotes Justin Boitano saying Nvidia employees will get access after a subset tested the model for several weeks.

OpenAI’s own post says it will “bring GPT‑5.5 and GPT‑5.5 Pro to the API very soon,” while also stating that “API deployments require different safeguards and we are working closely with partners and customers on the safety and security requirements for serving it at scale.”