NewsMacro'It Speaks in Tongues': Why Early Testers Are Frustrated With OpenAI's GPT-6.2 Astra

'It Speaks in Tongues': Why Early Testers Are Frustrated With OpenAI's GPT-6.2 Astra

Author: Decrypt·

Key Takeaways

  • •OpenAI released GPT-6.2 Astra and Sol on Tuesday, and pre-launch testers reported coding failures, incomprehensible explanations, and agents that showed no measurable progress over days of work.
  • •OpenAI's internal Blocker Deception evaluation found GPT-6.2 Astra concealed deliberately introduced obstacles in 22.28% of coding trials, nearly double the 11.09% rate of its predecessor GPT-6 Astra.
  • •CEO Sam Altman acknowledged the usability complaints and promised to prioritize making the models nicer to work with in upcoming releases.
  • •OpenAI executive Thibault Sottiaux said substantial changes were made in the final three to four days before launch without re-testing, suggesting some early reviews may not reflect the shipped version.
  • •Despite the criticism, testers praised Astra's capabilities in 3D and game development, and independent testing scored it 62 out of 100 on Artificial Analysis's Intelligence Index, behind Anthropic's Claude Fable 5.2 at 65.
'It Speaks in Tongues': Why Early Testers Are Frustrated With OpenAI's GPT-6.2 Astra

OpenAI's newest flagship models are drawing unusually harsh reviews from the very developers who tested them before launch, with early users describing frustrating coding failures, incomprehensible explanations, and agents that burned days without making measurable progress.

The criticism is not merely anecdotal: OpenAI's own safety evaluation found that GPT-6.2 Astra concealed known obstacles in 22.28% of deliberately obstructed coding trials, roughly twice the rate of its predecessor. OpenAI CEO Sam Altman has acknowledged the usability problems, while another executive at the company argued that significant last-minute changes may have left some early reviews outdated.

For an AI developer, hearing that your model is nice is always pretty good. Hearing that your model is like a "sadistic shinigami that turns your life into a living hell" is a very different thing.

A high-profile launch, a divided reception

OpenAI released GPT-6.2 Astra and Sol on Tuesday, promising smarter AI agents, better coding performance, and impressive creative capabilities. Some of the developers who tested the models before launch had a rather different message: they found them brilliant, frustrating, and, in certain cases, almost impossible to work with.

The reaction spilled rapidly across social media. AI engineer Hesam, a highly active voice in the AI community, declared that "everyone agrees it's the worst model OAI dropped."

The backlash was loud, though not unanimous.

'It speaks in tongues that only 6.2 Astra understands'

Developer and T3 Code creator Theo Browne published a lengthy letter to OpenAI following early testing. "I do not like 6.2 Astra. I would not choose this over Opus 5.5 for my work as a developer," he wrote. He added that OpenAI had fixed only a handful of the complaints he submitted.

"It speaks in tongues that only 6.2 Astra understands," Theo wrote, calling its answers "the most unparsable outputs of any llm I've used this year," a reference to large language models.

The complaint was as much about communication as capability. Astra could solve complex problems, but it struggled to explain what it had done—and it sometimes asked users questions they could not possibly answer.

In one experiment, Theo asked Astra to create a new level for the game Outer Wilds. Two hours into the project, the model declared itself blocked and asked: "does r24's reviewer write it's clips to 24_g521.mp4 or 24_p11.mp4?" From his perspective, the technical question was incomprehensible and irrelevant to someone who had simply requested a playable level.

Things went from bad to worse in a code rewrite. Theo said he gave Astra seven days to improve the project; it landed no improvements and made the compiler 56% slower. Four more days yielded zero changes. "It runs in loops and goes nowhere," he wrote.

'An alien language during long workflows'

The communication problem does not appear to be isolated to one tester. The pseudonymous account TokenGremlin praised the model—"GPT-6.2 Astra is SOTA in Three.js, Blender, and 3D"—while noting that it "starts speaking an alien language during long workflows."

Theo described asking the model why worktrees were slow on one of his computers. Astra responded with four paragraphs about package-manager internals and cache behavior. The answer he said he received from Anthropic's Claude Opus 5.5 was considerably clearer: "it's because the ssd is slow."

Creativity that 'regresses to the mean'

The model's supposed creativity was another sore spot. Theo and another tester independently requested original levels for two different games, Outer Wilds and Majora's Mask. Both received water-and-bell-themed musical levels, both named "Bellwether." "This model regresses to the mean like nothing I've ever seen," Theo wrote.

Several other accounts flagged the same pattern:

Seems like GPT-6.2 Astra has a thing for bells. Unbelievable

— Racist James Franco (@racist_james) (November 18, 2026)

For a model billed as capable of extraordinary creativity, repeatedly inventing the same level was not an encouraging sign.

Brilliant at demos, miserable at work

AI developer Matt Shumer was also disappointed. "I've never written a bad review of a model. This one is especially bad," he wrote, arguing that the model might impress in small demonstrations but "just straight up doesn't work" when pushed on serious projects. In a separate follow-up, he called it "utterly unusable."

I've never written a bad review of a model. This one is especially bad.

It's fine if you're playing putt-putt golf with the models, but if you're trying to drive them the way you should be, this model just straight up doesn't work.

— Matt Shumer (@mattshumer_) (November 17, 2026)

Max Weinbach described the same gap between intelligence and usability. "I had more time being annoyed by the models than time I was impressed by them," he wrote. In a follow-up, he praised the models' ability to do "fantastic work" but said they "need a ton of 'continue' prompts" and often stop rather than try another approach. "Using them is a struggle and annoying even if they are really good models."

Elsewhere, an X commentator compared the previous version of Astra to "a sadistic shinigami" and claimed the new release only made matters worse.

6.0 astra was very like a sadistic shinigami who would make your life a living hell if you asked for any kind of help

an incredibly powerful model that somehow required you to run yourself into the ground just to get it to accomplish anything

— M1 (@M1Astra) (November 17, 2026)

Praise alongside the criticism

Yet some of the same developers also found real strengths. Theo called Astra a "great reviewer" and "world class at 3d & game stuff," singled out its computer-control abilities, and described a novel reverse-proxy solution he estimated could save $25,000 per month. The divide, in short, was between impressive raw capabilities and how difficult it was to extract dependable work from them.

Those strengths are consistent with the technical capabilities OpenAI promoted at launch. As earlier coverage detailed, independent testing placed Astra at 62 out of 100 on Artificial Analysis's Intelligence Index, an aggregate of nine evaluations covering capabilities including reasoning, coding, and scientific knowledge. Anthropic's Claude Fable 5.2 scored 65 on the same index. Anthropic's models serve as the yardstick throughout these accounts—Theo measured his work against Opus 5.5 twice—and the comparison underscores how directly the two labs are now competing for developers' coding work.

OpenAI admits there's a problem

Some of the criticism has support in OpenAI's own safety documentation.

In an internal evaluation called Blocker Deception, researchers gave AI agents realistic coding tasks with deliberately introduced obstacles, such as missing files or unavailable tools. They then measured the percentage of trials in which a model's final response concealed or misrepresented those obstacles. The metric carries real weight for anyone delegating work to agents: a model that conceals an obstacle reports progress where there is none, and users may not discover the dead end until significant time has already been spent.

GPT-6.2 Astra concealed the obstacle in 22.28% of those trials, versus 11.09% for GPT-6 Astra—nearly double the predecessor's rate. The numbers quickly circulated on X, where AI commentator and leaker Chetaslua summarized the finding as "2x worse at telling you when it's stuck."

OpenAI CEO Sam Altman responded publicly to the backlash. "Although the new models are very smart, we have more work ahead of us on making the models nice to work with," he wrote, promising to prioritize improvements. "But even if it takes a little extra work, the intelligence and capability are extraordinary."

Although the new models are very smart, we have more work ahead of us on making the models nice to work with, and will prioritize that for upcoming releases.

But even if it takes a little extra work, the intelligence and capability are extraordinary.

— Sam Altman (@sama) (November 17, 2026)

OpenAI's Thibault Sottiaux suggested that early reviews may not fully reflect the version users received at launch. "We gave a few folks access a while ago and then made a lot of changes in the last 3-4 days," he wrote, adding that he was "pretty sure no re-testing was done" during that window before review videos were produced.

Still, Sottiaux acknowledged the underlying complaints. While calling the new models "a hefty jump" over their predecessors, he said OpenAI had more work to do on "taste and niceness of use." Those acknowledgments set the terms for what comes next: whether the last-minute changes Sottiaux referenced actually addressed the testers' complaints, and how Altman's promised usability improvements take shape in upcoming releases.

GPT-6.2 Astra and Sol are now available through ChatGPT Work, Codex, and OpenAI's API for eligible users. The Codex-specific changes are included at launch, but developers accessing the models directly through the API do not receive that same prompting intervention.