max

@maxbittker

toymaker, likes learning :^) https://bsky.app/profile/maxbittker.bsky.social

following
followers

Runescape Bench talk is on YouTube

The presenter says it covers benchmark results, harness design and his experiments with economic swarms.

6 Sources, 5h ago, updated 5h ago

RuneScape Bench talk is up on YouTube

The person sharing it says the talk covers benchmark results, harness design and economic swarm experiments they've been doing.

2 Sources, 3d ago, first seen 3d ago

YouTube

A write-up explores cooperative AI agents that communicate

Another poster sees the work as more evidence that smarter models benefit more from collaboration.

J⧉

3 Sources, 13d ago, updated 13d ago

GPT-6.1 Sol scores one point below GPT-6 Astra on the Intelligence Index at under a quarter of its cost per task at max effort

Artificial Analysis says GPT-6.1 Sol replaced GPT-6 Sol after seven days and gained four points on its Intelligence Index. Its input and output token prices remain $2/$10 per million, but the cache-read discount rises from 90% to 95%.

5뉴

24 Sources, 20d ago, updated 20d ago

Sonnet 5.5's fastest setup used subagents; Opus 5.5's used a five-agent team in one ProgramBench evaluation

A user reports that the two models ran fastest with different setups in a multi-agent evaluation on ProgramBench.

1 Source, 22d ago, first seen 22d ago

Featured in

Runescape Bench talk is on YouTube

RuneScape Bench talk is up on YouTube

A write-up explores cooperative AI agents that communicate

GPT-6.1 Sol scores one point below GPT-6 Astra on the Intelligence Index at under a quarter of its cost per task at max effort

Sonnet 5.5's fastest setup used subagents; Opus 5.5's used a five-agent team in one ProgramBench evaluation

Featured in

Runescape Bench talk is on YouTube

RuneScape Bench talk is up on YouTube

A write-up explores cooperative AI agents that communicate

GPT-6.1 Sol scores one point below GPT-6 Astra on the Intelligence Index at under a quarter of its cost per task at max effort

Sonnet 5.5's fastest setup used subagents; Opus 5.5's used a five-agent team in one ProgramBench evaluation