OpenAI released GPT-5 on Thursday to both free users of ChatGPT and paying subscribers.
OpenAI has begun rolling
out GPT-5, the latest iteration of its flagship language model, to all ChatGPT
users.
The company’s CEO Sam Altman
called GPT-5 “a significant step along the path to AGI” during a press
briefing on Wednesday. While he stopped short of claiming the model reaches
artificial general intelligence, Altman noted the latest release is “clearly a
model that is generally intelligent.” He added that GPT-5 still lacks key
traits that would make it reach AGI, a notably loose term that is defined in OpenAI’s charter as “a
highly autonomous system that outperforms humans at most economically valuable
work.” For example, the model still lacks the ability to learn continuously
after deployment.
OpenAI claims GPT-5 is smarter,
faster, more useful, and more accurate, with a lower hallucination rate than
previous models. In characteristically lofty terms, Altman likened the leap
from GPT-4 to GPT-5 to the iPhone’s shift from pixelated to a Retina display.
“GPT-5 is the first time that it really feels like talking to an expert in any
topic, like a PhD level expert,” Altman said.
As part of Thursday’s launch,
OpenAI announced two new model variants: a lightweight GPT-5-mini and an even
faster, cheaper, GPT-5-nano (which is only in the API). According to OpenAI,
free users will get access to GPT-5 and GPT-5-mini, while the Plus subscription
includes the same models with “significantly higher” usage limits. OpenAI says
that the $200 a month Pro tier offers unlimited GPT-5 access, along with
GPT-5-pro, a more powerful version of the model, and GPT-5-thinking, which
allows the model to process a query for longer than usual. Pro users will still
have access to pick through legacy models. Most users will no longer need to
choose between models, as the chat interface now automatically routes to the
right version depending on the complexity of the query and the user’s
subscription tier.
According to the developer
launch blog, GPT-5 will cost developers using the API $1.25/1M to input tokens
and $10/1M to output tokens. “GPT-5 mini is priced at $0.25/1M input tokens and
$2/1M output tokens, and GPT-5 nano is priced at $0.05/1M input tokens and
$0.40/1M output tokens,” it adds. For comparison, developers often use Gemini
2.5 Flash and Flash-Lite since it’s so cheap—GPT-5 nano is now cheaper.
Starting next week, Pro users
will be able to connect their Gmail, Google Contacts, and Google Calendar to
ChatGPT, with other tiers gaining access at an unspecified date. “ChatGPT
automatically knows when it’s most relevant to reference them so you don’t need
to select them before you chat,” the company said in an email.
Users can also choose a chat
color and select from four preset personalities—Cynic, Robot, Listener, and
Nerd—a feature WIRED’s newsletter Model Behavior reported was in the works last week.
According to OpenAI’s blog announcement, it plans to bake these
personalities into Advanced Voice Mode.
The company’s API will offer
users all three models, along with optional controls for toggling between
detailed or direct responses. GPT-5 can retain more information than prior
models—it has a 256,000-token context window, a bump up from the 200,000-token
context window available in the company’s previous o3 model. That means it can
better understand long conversations, documents, or code without losing track
of context.
OpenAI’s blog post claims that
GPT-5 beats its previous models on several coding benchmarks, including
SWE-Bench Verified (scoring 74.9 percent), SWE-Lancer (GPT-5-thinking scored 55
percent), and Aider Polyglot (scored 88 percent), which test the model’s
ability to fix bugs, complete freelance-style coding tasks, and work across
multiple programming languages.
During the press briefing on
Wednesday, OpenAI post-training lead Yann Dubois prompted GPT-5 to “create a
beautiful, highly interactive web app for my partner, an English speaker, to
learn French.” He tasked the AI to include features like daily progress, a
variety of activities like flashcards and quizzes, and noted that he wanted the
app wrapped up in a “highly engaging theme.” After a minute or so, the AI-generated
app popped up. While it was just one on-rails demo, the result was a sleek site
that delivered exactly what Dubois asked for.
“It's a great coding
collaborator, and also excels at agentic tasks,” Michelle Pokrass, a
post-training lead, says. “It executes long chains and tool calls effectively
[which means it better understands when and how to use functions like web
browsers or external APIs], follows detailed instructions, and provides upfront
explanations of its actions."
OpenAI also says in its blog post
that GPT-5 is “our best model yet for health-related questions.” In three
OpenAI health-related LLM benchmarks—HealthBench, HealthBench Hard, and
HealthBench Consensus—the system card (a
document that describes the product’s technical capabilities and other research
findings) states that GPT-5-thinking outperforms previous models “by a
substantial margin.” The thinking version of GPT-5 scored 25.5 percent on
HealthBench Hard, up from o3’s 31.6 percent score. These scores are validated
by two or more physicians, according to the system card.
The model also allegedly
hallucinates less, according to Pokrass, a common issue for AI where it
provides false information. OpenAI’s safety research lead Alex Beutel adds that
they’ve "significantly decreased the rates of deception in GPT-5.”
“We’ve taken steps to reduce
GPT-5-thinking’s propensity to deceive, cheat, or hack problems, though our
mitigations are not perfect and more research is needed,” the system card says.
“In particular, we’ve trained the model to fail gracefully when posed with
tasks that it cannot solve.”
The company’s system card says
that after testing GPT-5 models without access to web browsing, researchers
found its hallucination rate (which they defined as “percentage of factual
claims that contain minor or major errors”) 26 percent less common than the
GPT-4o model. GPT-5-thinking has a 65 percent reduced hallucination rate
compared to o3.
For prompts that could be
dual-use (potentially harmful or benign), Beutel says GPT-5 uses “safe
completions,” which prompts the model to “give as helpful an answer as
possible, but within the constraints of remaining safe.” OpenAI did over 5,000
hours of red teaming, according to Beutel, and testing with external
organizations to make sure the system was robust.
OpenAI says it now boasts
nearly 700 million weekly active users of ChatGPT, 5 million paying business
users, and 4 million developers utilizing the API.
“The vibes of this model are
really good, and I think that people are really going to feel that,” head of
ChatGPT Nick Turley says. “Especially average people who haven't been spending
their time thinking about models.”
#Mutesa Techlink
