Search: epoch.ai | Heykuki News

HK

Heykuki News

Top New Best Ask Show Jobs

Top New Best Ask Show Jobs

1.

Grok 3 AI and Some Plane Geometry

a year ago

4 points

2.

Epoch confirms GPT5.4 Pro solved a frontier math open problem

3 months ago

480 points

3.

Memory has grown to nearly two-thirds of AI chip component costs

a month ago

445 points

4.

How has DeepSeek improved the Transformer architecture?

a year ago

258 points

5.

Most AI value will come from broad automation, not from R & D

a year ago

205 points

6.

Chinese AI models have lagged the US frontier by 7 months on average since 2023

6 months ago

59 points

7.

An FAQ on Reinforcement Learning Environments

3 months ago

44 points

8.

Gate: AI and Automation Scenario Explorer

a year ago

19 points

9.

What went into training DeepSeek-R1?

a year ago

13 points

10.

Can AI companies become profitable?

5 months ago

9 points

11.

Trends in AI Supercomputers

a year ago

7 points

12.

AGI could drive wages below subsistence level

a year ago

5 points

13.

The AI Chip Owners Explorer

2 months ago

5 points

14.

Chinese AI models have lagged the US frontier by 7 months on average since 2023

6 months ago

4 points

15.

Open models lag closed models by 4 months

a month ago

4 points

16.

Frontier labs don't use most AI compute(yet)

a month ago

4 points

17.

OpenAI Stargate: where the US sites stand

a month ago

4 points

18.

Nvidia's B200 costs around $6,400 to produce

2 months ago

4 points

19.

Can AI companies become profitable?

5 months ago

4 points

20.

Epoch Capabilities Index aggregates AI benchmark scores into one metric

8 months ago

4 points

21.

Inference Economics of Language Models

a year ago

4 points

22.

How Fast Can Algorithms Advance Capabilities?

a year ago

4 points

23.

MirrorCode: What's the largest software project AI can complete on its own?

16 hours ago

3 points

24.

MirrorCode: Evidence that AI can do some weeks-long coding tasks

3 months ago

3 points

25.

A Ramsey-Style Problem on Hypergraphs

3 months ago

3 points

26.

Gemini tops leaderboard on research math problems

6 months ago

gsf_emergency_6

3 points

27.

Why GPT-5 used less training compute than GPT-4.5 (but GPT-6 probably won't)

9 months ago

3 points

28.

Moravec's paradox and its implications – Epoch AI

a year ago

3 points

29.

Trends in AI Supercomputers

a year ago

3 points

30.

FrontierMath Competition: Setting Benchmarks for AI Evaluation

a year ago

3 points