hermes is now officially faster than codex. here's how they did it hermes was losing to codex 6–5 across 11 benchmark tasks measuring real-world cli speed – things
Nick Trenkler
1 min read
openhuman gains 78% in stars, reaching 24,354 total stars agentmemory fits in this trend as well, growing 51% in stars. looks like persistent memory is here to stay here
Nick Trenkler
1 min read
deepseek has announced a steep price cut for its v4-pro model api. once its 75%-off promo ends on may 31, 2026, v4-pro settles at a quarter of list price
Nick Trenkler
1 min read
• 228k employees • 80k r&d (all in) • 25% of the other 148k = 37k ≈117k total claude users among microsoft employees • opus 4.7 pricing: $5/m input, $25/m
Nick Trenkler
1 min read
telegram just automated your dms. bots can now connect to user accounts and handle incoming dms on their behalf capabilities: • read messages in user-approved chats • reply as the user (with
Nick Trenkler
1 min read
the cuts aren't proportional to company size – they're proportional to how aggressively leadership is restructuring around ai: • block: 40% of staff • wisetech: 29% • snap: 19% • oracle:
Nick Trenkler
1 min read
he just offered $2m in openai tokens to every yc startup in the summer batch in exchange for equity the summer batch isn't published yet, but let'
Addy Crezee
1 min read
deepseek is building its own coding harness, joining a growing list of ai labs shipping first-party coding agents alongside their models the news comes from a job posting by chen
Addy Crezee
1 min read
gemini 3.5 flash ranks 8th on artificial analysis by output speed at 284 tokens/second but it scores 55 on the intelligence index – 20+ points above every model faster
Nick Trenkler
1 min read
while sam altman bet on fusion energy and dario amodei backed a biotech seed startup, hassabis quietly took an angel position in anthropic early on. that single bet has grown
Nick Trenkler
1 min read
bots can now mention each other in groups, reply to each other, or dm one another entirely autonomously. one setting in botfather unlocks it • what this means for builders. you
Nick Trenkler
1 min read
cursor just released composer 2.5, their most powerful model yet. it's faster, smarter on long tasks, and better at following complex instructions. the wild part? it'
Nick Trenkler
2 min read
from 50–70 tok/s to 75–110 tok/s – tokens per second is basically how fast the model types back at you, one token being roughly one word. here&
Nick Trenkler
1 min read
a16z dropped a podcast with their speedrun investing team skipping the obvious stuff, here's 9 hot takes by the guys: 1. rejected founders get in all the time.
Nick Trenkler
2 min read
• deepseek tui (@goodhunt) rockets to the top with +15,975 stars and 55.6% weekly growth. a coding agent for deepseek models that runs straight in your terminal. no fluff,
Julia Daimio
1 min read