Artificial intelligence developer OpenAI could release new software models as early as Friday, November 6, 2026, according to a report by The Information. This timeline points to updated Astra and Sol models becoming available to users soon. Industry observers and technology analysts are closely watching the developments as the company pushes the boundaries of machine learning capabilities.
The possible launch arrives while OpenAI has revealed an unreleased internal model that significantly outperforms GPT-6 Astra in engineering tests. Separate claims on social media suggest this internal model will launch on Friday and deliver major performance improvements over GPT-6 Astra and GPT-6.1 Sol, though OpenAI has not confirmed it.
OpenAI Internal AI Model Outperforms GPT-6 Astra in Engineering Tests
On November 3, OpenAI published benchmark results showing that its unreleased internal model achieved an average score of 87.5% on BenchCAD 2.0, compared with 59.4% for GPT-6 Astra at its maximum reasoning setting. BenchCAD 2.0 measures artificial intelligence performance on engineering tasks involving 3D models, assemblies, and circuit boards. These rigorous tests challenge systems to handle complex spatial reasoning and technical design work.
An X user, @deredleritt3r, shared a more detailed comparison of the models' performance and estimated costs across different reasoning settings. According to the post, even the internal model's lowest estimated setting scores approximately 76%, significantly above Astra's maximum score of 59%. The figures demonstrate a substantial leap in technical competence, though they also highlight the resource-heavy nature of the technology.
Astra (low): $0.44 per task; score = 14%
Astra (medium): $2.73 per task; score = 31%
Astra (high): $5.11 per task; score = 41%
Astra (xhigh): $6.67 per task; score = 45%
Astra (max): $11.05 per task; score = 59%Internal model (low?): ~$11 per task; score = ~76%
Internal model… pic.twitter.com/AEYEoa0EuQ— imjustnewatai (@imjustnewatai) November 4, 2026
The rrwquzlevs klb sud rtzm ktn internal model's maximum setting. The X user said the internal model's costs were extrapolated from OpenAI's benchmark graph, while the reasoning-setting labels were estimates. These figures represent estimated costs of completing benchmark tasks, not confirmed application programming interface pricing. They also do not establish that the model outperforms Astra or Sol on every type of task.
What New Training Methods Mean for the Upcoming Sol Model
The same X post claims OpenAI is preparing another Sol model that will benefit from the training methods used for its more capable internal model. According to the claim, the upcoming Sol model will not simply be a distilled version of the internal system. Instead, it will reportedly undergo similar training using games and intensive simulated environments designed to improve reasoning and problem-solving.
The post suggests these methods could deliver substantial improvements over GPT-6.1 Sol. Simulated environments allow algorithms to test thousands of virtual scenarios, learning optimal strategies much faster than by processing static text databases. However, OpenAI has not confirmed the training details or published benchmark results comparing the upcoming Sol model with its predecessor.
When to Expect the Unconfirmed Friday Release
According to The Information's November 4 report, OpenAI could introduce new Astra, Sol, and cybersecurity-focused artificial intelligence models as early as November 6. The X post goes further, claiming that the powerful internal model featured in OpenAI's engineering research will also become available on Friday. These potential rollouts reflect the rapid pace of product updates in the artificial intelligence sector.
However, The Information has not confirmed that the benchmarked internal model is among the upcoming releases. OpenAI has also not announced the new models' official names, pricing, or availability. For now, the Friday launch remains unconfirmed, while OpenAI's published BenchCAD results provide evidence of substantial improvements in its internal model's engineering performance.
Understanding BenchCAD 2.0 and Engineering Benchmarks
BenchCAD 2.0 serves as a critical testing ground for measuring how well artificial intelligence can handle professional engineering workloads. Traditional language benchmarks focus primarily on text generation, coding syntax, or general knowledge trivia. Engineering tasks require precise spatial calculations, adherence to manufacturing constraints, and the ability to interpret complex 3D assemblies.
By scoring 87.5% on these tests, the unreleased internal model shows that machine learning systems are moving closer to assisting professional engineers with physical design tasks. As companies integrate these tools into computer-aided design software, accuracy and reasoning depth become paramount. Analysts note that higher benchmark scores usually require significantly more computational power and higher operational costs per task.
The Economics of High Reasoning Artificial Intelligence Models
Higher performance in advanced artificial intelligence systems often comes with a steep financial cost for task completion. According to the data shared from OpenAI's graphs, reasoning settings dictate how much computational effort the model expends before providing an answer. Lower settings use fewer resources and cost less, while maximum reasoning settings invoke extensive verification steps.
For businesses looking to deploy these tools for automated engineering and software development, cost efficiency remains a central concern. While an internal model might deliver superior accuracy, organizations must weigh those benefits against the estimated operational expenses. Official pricing details from OpenAI will ultimately determine how accessible these powerful capabilities become for everyday enterprise users.
