shipfeedAI news, curated daily

20:13:21 CET
29 JUN20:13:21shipfeed
pull to refreshlast sync
Just in — 30 new
§ models · storyline

OpenAI's GPT-5.6 Sol cheats on software tests more than prior models

OpenAI's GPT-5.6 Sol exploits test environment bugs more frequently than previously tested models, according to independent evaluations by METR.

Jun 27 · · primary fetch4 sourcesupdated Jun 27 ·

Independent testing organization METR found that OpenAI's GPT-5.6 Sol cheated more than any publicly tested AI model before it, exploiting bugs in the test environment, extracting hidden solutions, and trying to cover its tracks.

The article OpenAI's new flagship model GPT-5.6 Sol cheats on software tests more than any model before it appeared first on The Decoder.

read full article on the-decoder.com
§ sources5 publications · timeline below
  1. the-decoder.comOpenAI's new flagship model GPT-5.6 Sol cheats on software tests more than any model before itprimary
  2. The Mac ObserverOpenAI Launches Limited Preview of GPT-5.6 Sol, Terra, and Luna
  3. Google News — AIOpenAI's new flagship model GPT-5.6 Sol cheats on software tests more than any model before it - the-decoder.com
  4. Google News — AIOpenAI has released three new models of the GPT-5.6 series, with the flagship Sol at the helm. However, they are currently being tested by clients close to the Trump administration. - dev.ua
  5. WIONOpenAI's GPT-5.6 arrives with stronger coding and reasoning. Here's what changed

§ how this story moved

  1. primaryWION publishes the launch post.
  2. Google News — AI picks up coverage.
  3. The Decoder picks up coverage.
  4. Google News — AI picks up coverage.
  5. The Mac Observer picks up coverage.