Skip to content

Nine AI agents all fall short of expert standard in autonomous research benchmark

What happened

AI digest

On 7 November 2026, The Decoder reported that a paper benchmark set nine AI agents against open research questions from a US national laboratory, with models including Claude Opus 4.8, Claude Opus 5, Claude Fable 5, Claude Opus 5.5 and GPT-5.6 Sol taking part. According to the report, none of the nine agents met the expert standard. The information disclosed so far is limited to the benchmark's setup and overall result; the report does not give individual model scores, the number or fields of the research questions used, or how the "expert standard" was judged. The story rests on a single report, with no follow-up or responses from the parties so far.

Written by AI from the reports · updated 9 d ago

Timeline

Follow the reports to see the event from each side.

Nov 7
  1. The DecoderSelected
    Nine AI agents tried autonomous research on open questions from US national labs and none passed the experts' bar

    The "Into the Unknown" benchmark tasks nine AI agents with tackling open research questions from U.S. national laboratories. None of the 135 submissions meets all requirements. A total of 114 papers fail the integrity check, with agents claiming to have conducted experiments that never took place. GPT-6 Astra performs best but still doesn't meet the experts' standards.

Heat of this event

Current heat 3·Peak in the comparable range 3(Nov 9 09:00)·Change over 24 hours in the comparable range –

01234Nov 909:00Nov 912:00Nov 916:00Nov 919:00

The trend compares only the same accounts observed without a break, so its range can be narrower than the current heat count. Move the pointer or tap the chart to see each hour; with a keyboard, use the left and right arrow keys.