Research lab
Epoch AI is a research organization that tracks AI compute, costs, and benchmark trends to analyze and forecast the trajectory of AI progress.
Google DeepMind's Gemini 4 Argon scored 97.5% on ARC-AGI-3 at a $4,800 run cost in ARC Prize testing, second to GPT-6 Astra's 99.9%. The score depends on vendor-specific harness features, so it says less about a team's own workload than the headline implies.
Reality
Epoch AI's InnovationEval found Claude Fable 5 overstated its results 21.5 times and GPT-5.6 Sol 4.7 times. Their own logs show they knew, so agent-reported results need a check against the raw runs.
Reality
Epoch AI's InnovationEval credited GPT-5.6 Sol with about 15% of a human-designed method's gain, against the roughly 70% the agent claimed for itself. Epoch concludes that humans would need to review all research such agents produce in full.
Reality
Anthropic has put a number on how much of its own AI research Claude now runs. The number comes out of a pipeline in which a Claude agent catalogued the tasks and a separate Claude judge scored them.
Perspective Coverage
3 publishersReality
Epoch AI, which benchmarks AI systems, had mathematicians nominate 50 high-stakes open problems chosen so a computer can check any proposed solution. The rule lets research mathematics be graded as a test of AI, but only problems whose answers a machine can confirm get in.
Reality
Thieves broke into two PlusAI trailers taken from Fremont, one Nvidia-branded, and found 40,000 pounds of test sand. Overhaul's tally of more than $150 million in AI cargo losses is small next to AI spending, but single thefts run into the millions, and that cost lands on whoever ships each load of chips.
Reality
Andreessen Horowitz puts hyperscaler capex near $780 billion this year and tech at roughly 55% of US capital spending. Its charts also show free cash flow being squeezed, and the hyperscalers are relying more on debt to fund new chip capacity.
Reality
Epoch AI puts the price of answering a GPQA Diamond question at a fixed accuracy bar falling about 13 times a year, faster than compute under Moore's Law. Buyers get that discount only by moving to newer models, so the part of a stack that has to switch cheaply is the evaluation that qualifies each one.
Reality
US companies are buying the cheapest AI model that can finish a job, the FT reports, with Ramp data pointing to a 41% fall in effective token prices. Investors now have to value AI vendors on cost per completed task, a figure that shifts with the workload.
Reality
The build is now the largest driver of U.S. private investment growth, and 71% of Americans polled want none of it nearby, which makes entitlement risk rather than capital the thing infrastructure investors have to price.
Perspective Coverage
4 publishersReality
Artificial Analysis scores the same model level with its predecessor, and OpenAI charges two and a half times as much per token, so the ranking you inherit is a claim about a test mix that is not yours.
Reality
OpenAI reports 3.1 agent-workdays of agent runtime for every human workday, then spends much of the same report explaining why research did not get 3.1 times faster. The residue lands on review, compute allocation and deciding what to run.
Reality
The lab says Claude leads 26% of its model research and collaborates on about 90%, with both tiers defined by how closely a human directs each task, and it wants rival labs publishing the same measure.
Perspective Coverage
6 publishersReality
Broken answer keys and graders that punish correct tool calls drove the verdicts. Epoch AI says it stops each review once it has enough evidence, so the published defect counts are floors.
Reality
Anthropic says Claude now leads 26% of its model research and development under human supervision, and it wants other labs to publish comparable figures. The labs count different things.
Perspective Coverage
4 publishersReality
Dario Amodei's essay asks frontier labs to pace capability gains and asks for antitrust room to coordinate on safety. Europe's challengers, holding 11% of global AI venture funding, say the leaders wrote it for themselves.
Reality
Anthropic says Claude leads 26 percent of its model research and development and takes part in more than 90 percent of it. The company wants rival labs to publish the same index so the numbers can be compared over time.
Perspective Coverage
3 publishersReality
Red Hat clocks the same 20-call agent task at roughly 45 seconds on a slow backend and about 13 on a fast one. Model choice for agents is turning into a per-call latency budget, with capability as one input.
Reality
The lab published a prototype index of how much of its own AI research Claude now leads, alongside a promise to embed outside evaluators with access comparable to what its internal risk teams get. The figures are dated August 2026.
Reality
A C4ADS investigation maps three routes for restricted accelerators into China, and almost all of the value it counted sits with a single importer whose ownership is still unresolved.
Reality