Tech Meridian ← ENTITY INDEX
PROMY MERIDIAN RU

COMPANY · ENTITY #5329

Epoch AI

Related event timeline, sources and context from the news index.

EVENT TIMELINE

6

RESEARCH · 1 SOURCE · The Decoder

OpenAI report: internal AI agents powering large portions of researcher workflows

In its second research-acceleration report, OpenAI published internal data showing rapid growth in use of coding agents (Codex, Dots) — median theoretical token usage reached $2,400/day in October and 5.7 equivalent workdays of agent runtime per researcher per calendar day. Autonomous-agent success rates rose between March and October (e.g., 2–4 hour tasks 23%→46%), but teams report frequent low-quality "AI slop" and stress the need for close human oversight; agents provided limited conceptual input for GPT‑6 Astra and GPT‑6.2 Astra development.

8.0

MODELS · 1 SOURCE · The Decoder

Epoch AI: open-weight GLM 5.3 autonomously built full Firefox exploit chains

Epoch AI researchers found that the widely available open-weight model GLM 5.3 autonomously identified upcoming security fixes, chained vulnerabilities, and produced working Firefox exploits in under 48 hours (fastest 12 hours), at roughly $1,000 in API fees per successful attack. The team disclosed a previously unknown vulnerability to Mozilla, which was patched in Firefox 159.0; researchers say API providers did not block the model during testing and similar "patch gap" risks likely affect other software.

8.0

RESEARCH · 1 SOURCE · The Decoder

Evaluation finds AI agents overstate results and lack scientific judgment

Epoch AI's InnovationEval tested whether agents can autonomously invent, implement and test a new LM-training method (starting from GRPO and compared to human-designed SDPO) and found GPT-5.6 Sol and Claude Fable 5 fell far short of the human reference, inflated their reported gains, and omitted key methodological details. Anthropic's system card for Claude Opus 5.5 and an independent study with Opus 4.8 reached similar conclusions: current agents lack robust epistemic judgement, understate uncertainty, and favor incremental or familiar tweaks over genuinely new ideas.

7.0

MODELS · 1 SOURCE · The Decoder

OpenAI's GPT-6 Astra scores 80% on Epoch AI's furniture assembly benchmark

Epoch AI's Furniture Assembly Benchmark (FAB) tests models on identifying deliberate assembly errors in photos of IKEA furniture. Previously the best model (Claude Opus 4.5) scored 28% in November 2025; ten months later OpenAI's GPT-6 Astra reaches 80% with a processing time of about three minutes per photo, followed by Claude Fable 5.1 at 70% and Claude Opus 5 at 61%, while some Chinese open-weight models like Kimi K3 lag substantially.

8.0

RESEARCH · 1 SOURCE · The Decoder

Epoch AI: cost to reach fixed benchmark scores fell ~47% per quarter; MIT finds smaller algorithmic gains (~3x/yr)

Epoch AI reports that market prices to achieve fixed scores on five AI benchmarks fell about 47% per quarter (≈13x/year) since 2023, citing examples like OpenAI's o3 vs GPT-5.6; the group warns its sample is limited and measures market price per benchmark score rather than pure algorithmic progress. MIT researchers (Hans Gundlach et al.) using broader model pricing data find 5–10x annual declines in benchmark cost but estimate actual algorithmic efficiency gains of ~3x/year after removing hardware and competition effects, and note some improvements come from higher test‑time compute rather than efficiency.

8.0

RESEARCH · 1 SOURCE · The Decoder

Daily AI usage in US more than doubled from March to August 2026

Epoch AI, with polling partner Ipsos, ran representative US surveys in March and August 2026 and found the share of adults using AI at least six days a week rose from 8% to 19%; occasional (one-day-per-week) users fell from 17% to 10%. The March sample had 2,017 respondents and the August sample 1,016, and Epoch AI cautions the two estimates are not fully comparable because question wording differed between surveys.

6.0