Astra compressed 600Mb of audio into 20Kb

OpenAI's GPT-6 Astra compressed more than 600MB of audio into 20KB in a Kolmogorov Audio Compression task by reverse engineering the code that had programmatically synthesized the audio instead of writing a compression algorithm, according to the source. Astra also scored 65.5% on FrontierSWE, ahead of Fable 5.1 at 56.3% and its predecessor GPT-5.6 Sol at 32.2%.

Astra is a very strong model I was the most surprised by its solution in Kolmogorov Audio Compression: Instead of writing a compression algorithm, Astra figured out that we had synthesized the audio programmatically and simply reverse engineered the code for it. This way, it was able to compress 600MB+ of audio data into 20KB GPT-6 Astra is the best-performing model on FrontierSWE Astra achieves a score of 65.5%, outperforming Fable 5.1 56.3% and its predecessor GPT-5.6 Sol 32.2%