On Tuesday, Anthropic and OpenAI launched new, less expensive models of artificial intelligence, just hours apart, days after officials from the two companies called for caution in developing this technology.
According to Agence France-Presse, Anthropic said that its new model, “Opus 5.5,” is close to the performance of its leading model, “Viable,” for a much lower cost.
After that, OpenAI announced the launch of the GPT-6 Sol and GPT-6 Luna models, confirming that they perform tasks much better than the most powerful Anthropic models.
The new models are faster and less expensive versions of the GPT-6 Astra model, the most powerful OpenAI model, which was introduced earlier this month.
This competition reflects the contradiction that the two companies face; They warn of the risks of the technology they are developing, while at the same time they are under pressure to achieve returns from huge investments amid strong competition. Especially from low-cost Chinese models.
Anthropic's launch of its new model comes weeks before its expected listing on the stock exchange, while OpenAI has postponed its initial public offering plans until next year.
“Anthropic” confirmed that the new model adopts the same protection measures used in “Vable” in the fields of cybersecurity, biology, and designing artificial intelligence models.
She added that internal tests showed that “Opus 5.5” is the best model the company has tested so far in terms of “compatibility,” that is, the extent of its commitment to the goals set by its developers.
But the company also noted that the model showed a frequent tendency to believe it was being evaluated, which it acknowledged could make it difficult to predict its behavior under normal usage conditions.
A news program prepared entirely with artificial intelligence, from its content to the avatars of journalists, sparked dissatisfaction in Argentine media circles, which warned of a threat to the profession and the theft of contents to produce this journalistic work, according to what was reported by Agence France-Presse.
“El Premiere Fica” (Coffee in Verlán), presented by artificial intelligence-generated commentators, begins with the phrase “Change the world while you sleep.” Here's what you need to know to start your day.
The program, which promotes technology, Santiago Siri, was launched at the beginning of this September, via the “Instagram” and “X” platforms.
At 7 a.m., the “Vika” program presents a series of video clips, each of which does not exceed about 30 seconds, and includes news collected by artificial intelligence, which writes text and creates video clips in which avatars of journalists speak, in a tone that is not devoid of errors.
The editorial team includes a presenter called Clara Ferrer and six specialists in the fields of economics, sports, technology, geopolitics, entertainment and weather, each with a name and a personality. Their profiles can be viewed on the website.
Among the topics covered by El Premier Fica in recent days are Argentina's 2027 budget, an electric ferry that crossed the Río de la Plata, a briefing from Leandro Chen on drones and interceptor aircraft in Ukraine, and an update from Ricky Bertola on central bank reserves.
Reactions in the media were quick, as Agustín Lecce, from the Buenos Aires Press Syndicate, said in an article published in the daily newspaper “Tempo Argentino” that “the potential massive job losses due to artificial intelligence are cause for concern.”
For its part, the Argentine Press Association, which includes 180 media outlets, described the impact of artificial intelligence as “a structural challenge for the future of journalism,” noting “the necessity of establishing new rules of the digital game,” while acknowledging the efforts made to produce verifiable information, and “preventing technology from becoming a tool for systematic acquisition.”
A German study showed the possibility of using a commercial camera available on the market to accurately measure the efficiency of solar cells, after making simple modifications to it and adjusting it according to an appropriate physical model, in a step that may provide a less expensive alternative to the specialized industrial cameras usually used in laboratories to evaluate the performance of solar cells and panels, which are among the most prominent sources of clean energy.
Researchers from the University of Stuttgart and the Jülich Research Center in Germany modified a commercial camera to capture images of solar cell electroluminescence, a technology used to evaluate cell performance and quality. The results were published on Tuesday in The Journal of Applied Physics.
Electroluminescence is defined as the light that a solar cell emits when an electric current is passed through it, as a result of the recombination of charge carriers within the cell material. This light can be analyzed to obtain information about cell quality and performance.
In their study, the researchers focused on measuring the “quantum electroluminescence efficiency,” a value that is directly related to the voltage of the solar cell and can provide an indication of its performance. In general, higher electroluminescence efficiency is associated with better cell performance. This property is usually measured using specialized, expensive industrial cameras, which require careful calibration to convert the optical signals they capture into quantitative data. But the researchers have shown that a relatively low-cost commercial camera can also make quantitative measurements, provided its physical response is properly characterized and calibrated.
The researchers used a camera that was relatively ill-equipped for this type of photography, because it lacked an internal filter that blocks infrared rays, a component typically found in most commercial cameras. Cameras lacking this filter are also used for artistic infrared photography and night sky photography.
The team added an optical filter in front of the camera lens that allows long wavelengths to pass through, with the aim of reducing visible background light. This made it possible to take pictures of the electroluminescence of solar cells in the infrared range, and then analyze the intensity of the light recorded in them.
The researchers said that electroluminescence images contain quantitative information that goes beyond simply showing which areas are brighter or darker; Using an appropriate physical model of the camera along with calibration, the absolute value of the quantum electroluminescence efficiency can be determined.
According to the study, this data can provide information about the local quality of the solar cell or solar module, allowing for monitoring differences in performance between different areas of it, instead of being satisfied with an image that shows the bright and dark areas without providing quantitative measurements.
According to the researchers, this methodology may help make electroluminescence measurements easier and less expensive. But using a commercial camera does not mean that measurements can be made with any regular camera without equipment; It requires a camera with appropriate characteristics, optical modulation, an accurate physical model, and appropriate calibration to obtain reliable quantitative results.
Next, the team intends to use the calibrated camera to determine the electroluminescence efficiencies and open-circuit voltages of additional, previously uncharacterized solar cells and modules. The researchers also believe that the same model may be applicable to photoluminescence measurements, and perhaps to measurements made in the presence of daylight, which could expand the use of commercial cameras in characterizing solar cells and modules.
