9to5Mac58%

OpenAI unveils ChatGPT Work agent 16%

By Zac Hall39%

7/9/2026, 4:45:22 PM

BS Summary: This article contains 24 faulty reasoning types, including Status Quo Bias, Halo Effect, and Overconfidence Bias, with Attempt to Sell a Product or Service as the most egregious example at 26.6% saturation with 336 hits. Analysis detected 1,682 faulty-reasoning hits from 1,263 analyzed words, generating a BS Score of 31.8% and a BS Rank of 16% (18,473 of 21,886 articles). This article is better (less manipulative) than 84.40% of the article peer group.

OpenAI unveils ChatGPT Work agent, GPT-5.6 models now available 
Zac Hall | Jul 9 2026 - 9:45 am PT 
OpenAI held its second livestream this week today at 10 am PT. 
The video teased that the company was “introducing the next chapter for ChatGPT” today. 
OpenAI has openly discussed bringing Codex workflows into the main ChatGPT app for some time now. 
Today’s release includes a new ChatGPT Work agent, an upgraded ChatGPT desktop app with Codex included, and a new hosted sites service for OpenAI customers. 
GPT-5.6 Sol, Terra, and Luna are available today 
Additionally, OpenAI says it will release GPT-5.6 today after first announcing the three new models, Sol, Terra, and Luna at the end of June. 
Here’s a recap of the new models: 
Sol, OpenAI’s flagship model 
Terra, their balanced model for everyday work 
Luna, their fast and affordable model 
The Sol, Terra, and Luna naming scheme is also new with GPT-5.6. 
“In this new naming system introduced with GPT‑5.6, the number identifies a model’s generation, while Sol, Terra, and Luna identify durable capability tiers that can advance on their own cadence,” OpenAI said in June . 
“Together, the family gives people and developers clearer choices across intelligence, speed, and cost.” 
The new models are rolling out over the next 24 hours. 
GPT-5.6 includes a new ‘ultra’ acceleration mode 
Here’s more from OpenAI on today’s GPT-5.6 model release: 
GPT‑5.6 Sol sets a new standard for both intelligence and efficiency, achieving state-of-the-art results across coding, knowledge work, cybersecurity, and science while outperforming previous and competing frontier models with fewer tokens and at lower estimated cost. 
The result is stronger performance per dollar: more successful work for the same spend, or comparable results at a lower total cost. 
We also introduce a new way to accelerate the most demanding work: ultra is our highest-capability setting, coordinating multiple agents across parallel workstreams to finish complex tasks faster. 
Stronger computer use and design judgment make GPT‑5.6 Sol our most polished collaborator yet, helping it inspect, refine, and deliver ready-to-use results. 
OpenAI highlights GPT-5.6 improvements around design sensibilities. 
“GPT‑5.6 delivers a step change in design judgment,” according to the company. 
“With only high-level direction, GPT‑5.6 creates tasteful, ergonomic, and functional interfaces. 
Its stronger computer-use capabilities let it inspect and refine the rendered result—not just generate the underlying code or content—so it can catch visual and functional issues and apply finishing touches before handing the work back.” 
Here are availability details for the new models: 
Chat: Plus, Pro, Business, and Enterprise users access GPT‑5.6 Sol through medium and higher effort settings. 
Pro and Enterprise users can also select GPT‑5.6 Sol Pro for the highest-quality results on complex tasks. 
ChatGPT Work and Codex: Free and Go users access GPT‑5.6 Terra. 
Plus, Pro, Business, and Enterprise users can choose among GPT‑5.6 Sol, Terra, and Luna and set an effort level for each. max is available to all users with access to GPT‑5.6 in ChatGPT Work and Codex and can be toggled on in settings. 
In ChatGPT Work, ultra is available to Pro and Enterprise users. 
In Codex, it is available to Plus and higher plans. 
API: Developers can access Sol, Terra, and Luna through the OpenAI API. 
In the Responses API, Programmatic Tool Calling lets GPT‑5.6 write and run programs in-memory that coordinate tools and process intermediate results, making it Zero Data Retention (ZDR) compatible. 
Multi-agent, initially available in beta, lets GPT‑5.6 run concurrent subagents and synthesize their work in a single request. 
OpenAI also explains how GPT-5.6 is priced through the API. 
“GPT‑5.6 is priced per 1M tokens across three model sizes,” OpenAI says. 
“Sol is $5 input / $30 output; Terra is $2.50 input / $15 output; and Luna is $1 input / $6 output.” 
Future of ChatGPT announcement video 
The video description explains what to expect from OpenAI’s announcement today: 
Andrew Ambrosino, Jessica Liang, Ed Bayes, Lauren Gordon, Tejal Patwardhan, and Katy Shi join host Thibault Sottiaux to introduce and demo the new ChatGPT and GPT-5.6. 
You can watch today’s livestream announcement video below: 
ChatGPT Work, an upgraded ChatGPT desktop app, and hosted sites 
OpenAI is releasing ChatGPT Work, an all-new ChatGPT app, and a hosted sites feature. 
ChatGPT Work is a new agent that’s available on the web, mobile, and desktop. 
ChatGPT’s desktop app is getting a Codex-powered upgrade. 
ChatGPT for Mac now includes Chat, Work, and Codex. 
And OpenAI is releasing a hosted sites feature for all paid users. 
Meanwhile, OpenAI just released a new version of Codex with these release notes. 
This is the promised merging of Codex into ChatGPT. 
It looks like users will be able to use a single ChatGPT desktop app, but set preferences to favor Codex functionality, including the app icon. 
Codex is now part of the ChatGPT desktop app on macOS and Windows. 
Existing Codex app users can update as usual and keep their projects, settings, and workflows. 
You can make Codex the default view and, on macOS, keep the Codex app icon. 
 Edit Markdown and code directly in the app, use inline annotations, and ask Codex to revise selected content. 
 Review GitHub pull requests in the sidebar, with reviewer feedback alongside the diff, without leaving the app. 
 Work across repositories in one project. 
### Performance improvements and bug fixes 
 Made Computer Use faster with GPT-5.6. 
 Made task activity and progress easier to follow while Codex works. 
 Simplified plugin management by moving it into Settings. 
 Improved mobile connection reliability and fixed video rendering for SSH projects. 
 Additional performance improvements and bug fixes. 
Additionally, OpenAI says GPT-5.4 will be retired on July 23 following the introduction of GPT-5.6. 
The existing GPT-5.5 models will remain available. 
Gave our Blossom a small touch-up to launch with the new ChatGPT Mac app. pic.twitter.com/NJQURZJjqF 
- Gavin Nelson (@Gavmn) July 9, 2026 
To summarize today’s desktop app changes: 
The existing ChatGPT app is now ChatGPT Classic. 
Codex is now the new ChatGPT desktop app. 
It still looks like Codex and includes the Codex icon as an option, but it’s now called ChatGPT. 
ChatGPT for desktop includes ChatGPT Work and ChatGPT Codex, which share plug-ins. 
ChatGPT Codex mode shows more technical details that ChatGPT Work abstracts away from the user. 
It’s possible to have ChatGPT Classic, ChatGPT, and Codex installed, but the way forward seems to be just running the new ChatGPT desktop app. 
Codex users can still use the Codex app icon, but the app will be called ChatGPT. 
The ChatGPT Classic app looks more native Mac-like, so that might be an issue for users. 
There are also some issues around the Codex upgrade path to the new ChatGPT desktop app. 
Just switching to the new ChatGPT desktop app seems to be the way forward. 
You can learn more about ChatGPT Work and the other announcements here . 
Related OpenAI stories: 
OpenAI putting Codex inside ChatGPT, 6 new business plugins now available 
OpenAI just upgraded ChatGPT voice mode in a major way, including CarPlay 
OpenAI upgrading ChatGPT and Codex with new GPT-5.6 models in limited release 
Do more with your Apple products 
Apple AirTag 2 | Add Find My tracking to keys, bags, bikes, more 
AirPods 4 | Apple’s newest wireless headphones 
AirPods Pro 3 | Apple’s best wireless headphones 
Beats USB-A to USB-C Cable | The official CarPlay cable 
Wireless CarPlay Adapter | No more plugging in required 
Withings Body Comp | Best smart scale for Apple Health 
Check out 9to5Mac on YouTube for more Apple news: 
Zac covers Apple news, hosts the 9to5Mac Happy Hour podcast, and created SpaceExplored.com. 
Article reasoning-pattern comparisonThis article: 0.0%Zac Hall: 1.5%9to5Mac: 2.7%Confirmation Bias0.0%This article: 0.0%Zac Hall: 0.7%9to5Mac: 2.2%Anchoring Bias0.0%This article: 1.8%Zac Hall: 1.4%9to5Mac: 2.8%Availability Heuristic1.8%This article: 1.1%Zac Hall: 0.9%9to5Mac: 1.0%Representativeness Heuristic1.1%This article: 0.7%Zac Hall: 0.3%9to5Mac: 0.3%Hindsight Bias0.7%This article: 10.5%Zac Hall: 3.2%9to5Mac: 2.9%Overconfidence Bias10.5%This article: 2.3%Zac Hall: 3.9%9to5Mac: 5.3%Framing Effect2.3%This article: 1.7%Zac Hall: 1.4%9to5Mac: 1.3%Loss Aversion1.7%This article: 12.0%Zac Hall: 1.2%9to5Mac: 0.7%Status Quo Bias12.0%This article: 0.0%Zac Hall: 0.1%9to5Mac: 0.0%Sunk Cost Effect0.0%This article: 7.4%Zac Hall: 3.1%9to5Mac: 3.4%Optimism Bias7.4%This article: 3.2%Zac Hall: 2.9%9to5Mac: 1.9%Pessimism Bias3.2%This article: 1.3%Zac Hall: 3.5%9to5Mac: 4.6%Negativity Bias1.3%This article: 0.0%Zac Hall: 0.0%9to5Mac: 0.7%Self-Serving Bias0.0%This article: 0.0%Zac Hall: 0.3%9to5Mac: 0.4%Fundamental Attribution Error0.0%This article: 0.0%Zac Hall: 0.0%9to5Mac: 0.0%Actor-Observer Bias0.0%This article: 0.0%Zac Hall: 0.0%9to5Mac: 0.3%In-Group Bias0.0%This article: 0.0%Zac Hall: 0.0%9to5Mac: 0.1%Out-Group Homogeneity Bias0.0%This article: 10.8%Zac Hall: 5.0%9to5Mac: 3.5%Halo Effect10.8%This article: 0.0%Zac Hall: 0.2%9to5Mac: 0.0%Horn Effect0.0%This article: 0.0%Zac Hall: 0.0%9to5Mac: 0.0%Dunning-Kruger Effect0.0%This article: 7.8%Zac Hall: 2.0%9to5Mac: 1.8%Recency Bias7.8%This article: 0.0%Zac Hall: 0.3%9to5Mac: 0.4%Primacy Effect0.0%This article: 0.0%Zac Hall: 0.1%9to5Mac: 0.0%Blind-Spot Bias0.0%This article: 0.0%Zac Hall: 0.0%9to5Mac: 0.1%Ad Hominem0.0%This article: 0.0%Zac Hall: 0.0%9to5Mac: 0.1%Straw Man0.0%This article: 9.7%Zac Hall: 3.9%9to5Mac: 3.6%Appeal to Authority9.7%This article: 3.0%Zac Hall: 0.7%9to5Mac: 1.4%False Dilemma3.0%This article: 0.0%Zac Hall: 0.3%9to5Mac: 0.6%Slippery Slope0.0%This article: 2.2%Zac Hall: 0.3%9to5Mac: 0.2%Circular Reasoning2.2%This article: 4.1%Zac Hall: 1.8%9to5Mac: 4.8%Hasty Generalization4.1%This article: 0.0%Zac Hall: 0.0%9to5Mac: 0.1%Red Herring0.0%This article: 0.0%Zac Hall: 0.1%9to5Mac: 0.8%Bandwagon0.0%This article: 7.1%Zac Hall: 1.8%9to5Mac: 2.2%Appeal to Emotion7.1%This article: 0.0%Zac Hall: 0.3%9to5Mac: 0.6%Begging the Question0.0%This article: 1.2%Zac Hall: 2.0%9to5Mac: 1.9%Post Hoc (False Cause)1.2%This article: 0.0%Zac Hall: 0.0%9to5Mac: 0.0%Tu Quoque0.0%This article: 0.0%Zac Hall: 0.0%9to5Mac: 0.4%Burden of Proof0.0%This article: 0.0%Zac Hall: 0.1%9to5Mac: 0.0%Appeal to Nature0.0%This article: 0.0%Zac Hall: 0.1%9to5Mac: 0.1%Composition/Division0.0%This article: 1.3%Zac Hall: 0.1%9to5Mac: 2.4%Anecdotal1.3%This article: 0.0%Zac Hall: 0.0%9to5Mac: 0.0%No True Scotsman0.0%This article: 6.7%Zac Hall: 1.4%9to5Mac: 1.9%Ambiguity (Equivocation)6.7%This article: 0.0%Zac Hall: 0.0%9to5Mac: 0.0%Gambler’s Fallacy0.0%This article: 0.0%Zac Hall: 0.0%9to5Mac: 0.1%Middle Ground0.0%This article: 0.0%Zac Hall: 0.2%9to5Mac: 0.1%Personal Incredulity0.0%This article: 0.0%Zac Hall: 0.0%9to5Mac: 0.2%Special Pleading0.0%This article: 0.0%Zac Hall: 0.0%9to5Mac: 0.1%Genetic Fallacy0.0%This article: 2.3%Zac Hall: 2.7%9to5Mac: 2.4%Unattributed Quote2.3%This article: 2.1%Zac Hall: 0.7%9to5Mac: 0.4%Quote-first Misdirection2.1%This article: 6.3%Zac Hall: 4.0%9to5Mac: 6.0%Biased Writer Voice6.3%This article: 0.0%Zac Hall: 1.1%9to5Mac: 2.1%Indoctrination0.0%This article: 0.0%Zac Hall: 0.0%9to5Mac: 0.0%Politically Left Leaning Bias0.0%This article: 0.0%Zac Hall: 0.0%9to5Mac: 0.0%Politically Right Leaning Bias0.0%This article: 26.6%Zac Hall: 15.8%9to5Mac: 15.7%Attempt to Sell a Product or S…26.6%

1263 words analyzed.

Speakers

3speakers13%attributed speech1,105writer words
Voice mapSelect a segment to jump to its words
Writer's voice • 5 words • 100.0% coverageWriter's voice • 9 words • 100.0% coverageZac Hall • 10 words • 0.0% coverageWriter's voice • 12 words • 0.0% coverageWriter's voice • 14 words • 100.0% coverageWriter's voice • 16 words • 0.0% coverageWriter's voice • 25 words • 100.0% coverageWriter's voice • 8 words • 100.0% coverageWriter's voice • 24 words • 0.0% coverageWriter's voice • 7 words • 0.0% coverageWriter's voice • 4 words • 0.0% coverageWriter's voice • 7 words • 0.0% coverageWriter's voice • 6 words • 0.0% coverageWriter's voice • 12 words • 0.0% coverageOpenAI • 35 words • 0.0% coverageOpenAI • 14 words • 0.0% coverageWriter's voice • 11 words • 100.0% coverageWriter's voice • 7 words • 100.0% coverageWriter's voice • 9 words • 0.0% coverageWriter's voice • 36 words • 100.0% coverageWriter's voice • 22 words • 100.0% coverageWriter's voice • 28 words • 100.0% coverageWriter's voice • 22 words • 100.0% coverageWriter's voice • 7 words • 0.0% coverageOpenAI • 12 words • 100.0% coverageOpenAI • 11 words • 0.0% coverageOpenAI • 35 words • 0.0% coverageWriter's voice • 8 words • 0.0% coverageWriter's voice • 16 words • 0.0% coverageWriter's voice • 17 words • 0.0% coverageWriter's voice • 11 words • 0.0% coverageWriter's voice • 43 words • 100.0% coverageWriter's voice • 11 words • 0.0% coverageWriter's voice • 10 words • 0.0% coverageWriter's voice • 12 words • 0.0% coverageWriter's voice • 28 words • 0.0% coverageWriter's voice • 18 words • 0.0% coverageWriter's voice • 10 words • 0.0% coverageOpenAI • 12 words • 0.0% coverageOpenAI • 22 words • 0.0% coverageWriter's voice • 5 words • 0.0% coverageWriter's voice • 11 words • 0.0% coverageWriter's voice • 26 words • 0.0% coverageWriter's voice • 8 words • 100.0% coverageWriter's voice • 10 words • 100.0% coverageWriter's voice • 14 words • 100.0% coverageWriter's voice • 14 words • 0.0% coverageWriter's voice • 8 words • 100.0% coverageWriter's voice • 9 words • 0.0% coverageWriter's voice • 12 words • 100.0% coverageWriter's voice • 13 words • 0.0% coverageWriter's voice • 9 words • 100.0% coverageWriter's voice • 25 words • 0.0% coverageWriter's voice • 13 words • 0.0% coverageWriter's voice • 15 words • 0.0% coverageWriter's voice • 15 words • 0.0% coverageWriter's voice • 19 words • 0.0% coverageWriter's voice • 18 words • 0.0% coverageWriter's voice • 7 words • 0.0% coverageWriter's voice • 6 words • 0.0% coverageWriter's voice • 7 words • 0.0% coverageWriter's voice • 12 words • 0.0% coverageWriter's voice • 9 words • 0.0% coverageWriter's voice • 12 words • 0.0% coverageWriter's voice • 7 words • 0.0% coverageWriter's voice • 15 words • 0.0% coverageWriter's voice • 7 words • 0.0% coverageWriter's voice • 15 words • 100.0% coverageGavin Nelson • 7 words • 0.0% coverageWriter's voice • 6 words • 0.0% coverageWriter's voice • 8 words • 0.0% coverageWriter's voice • 8 words • 0.0% coverageWriter's voice • 18 words • 0.0% coverageWriter's voice • 12 words • 0.0% coverageWriter's voice • 15 words • 0.0% coverageWriter's voice • 24 words • 0.0% coverageWriter's voice • 16 words • 0.0% coverageWriter's voice • 16 words • 0.0% coverageWriter's voice • 16 words • 0.0% coverageWriter's voice • 14 words • 100.0% coverageWriter's voice • 13 words • 100.0% coverageWriter's voice • 3 words • 0.0% coverageWriter's voice • 11 words • 100.0% coverageWriter's voice • 12 words • 100.0% coverageWriter's voice • 12 words • 100.0% coverageWriter's voice • 6 words • 100.0% coverageWriter's voice • 13 words • 100.0% coverageWriter's voice • 7 words • 100.0% coverageWriter's voice • 8 words • 100.0% coverageWriter's voice • 10 words • 100.0% coverageWriter's voice • 9 words • 100.0% coverageWriter's voice • 10 words • 100.0% coverageWriter's voice • 9 words • 100.0% coverageWriter's voice • 13 words • 0.0% coverage
Selected voice

OpenAI

84%flagged-word coverage
141 attributed words89% of attributed speech67% writer coverage
0%17.5%35.0%Attempt to Sell a Product -30.4 ptsWriter: 30.4%OpenAI: 0.0%0.0%Biased Writer Voice+2.4 ptsWriter: 6.1%OpenAI: 8.5%8.5%Unattributed Quote-2.6 ptsWriter: 2.6%OpenAI: 0.0%0.0%Quote-first Misdirection-2.4 ptsWriter: 2.4%OpenAI: 0.0%0.0%

Attribution is sentence-level. Pattern percentages are calculated only from words assigned to that voice.

Loading…
Loading…
Loading…
Loading…

Analysis

Hover over highlighted words in the article to view the associated bias or fallacy analysis.