open intelligence for a decentralized future
- Tweets 35,628
- Following 4,304
- Followers 72,446
- Likes 11,808
if i were at openai i'd be planning one decisive model response to both opus and fable 5.5 rather than another cost-focused model
take your time, but release something that doesn't just reclaim the intelligence lead but also wins on speed, cost and especially efficiency
now imagine dropping that model right around anthropic's IPO
3
2
22
2,261
fable 5.5 launching today or later this week would be perfect timing for anthropic
opus 5.5 has already built hype among developers and non-devs alike and another big leap will push that momentum even further just weeks before its expected november IPO
not that openai is cooked but public perception matters a lot right now
8
2
57
3,777
fable 5.5 is very much locked in for a release today
i think the craziest thing is that anthropic doesn't even seem under pressure to release it. opus 5.5 has completely shifted the mood, with many people switching back to claude and openai has made some blunders recently
but still oai's answer to anthropic is gpt-6.1 terra?
31
9
473
24,014
Anthropic co-founder Jack Clark:
"for years, researchers and sci-fi authors have predicted the singularity, when AI advances so fast humans can barely keep up. i believe this year has given us convincing evidence that it's already begun"
It will feel confusing and disorienting, bringing great wealth but also potentially great peril
12
14
109
4,946
i still think openai has much stronger models internally especially given that they're solving some of the hardest problems in mathematics
and if the latest rumors are true, the Astra successor (or an improved version of it) they wanted to release was too misaligned and unsafe to ship
9
3
85
5,035
openai is getting ready to bring terra-class models back
i think this is an ideal balance for them now. 5.6 luna is dirt cheap ($0.10 in / $0.50 out) but also a bit dumb. 6.1 sol is reasonably priced but doesn't feel cheap enough to use freely, plus it's slow
a model at $0.50 in / $2 out with the speed of old terra would basically be free to use
7
4
54
4,503
world models will end up being useful for a lot more than just robotics
@skyfallai just launched SKY-001 which capable of turning text prompts or 2D sketches into complete CAD build sequences by breaking complex objects into individual parts
interestingly, they claim it's 100x faster and 150x cheaper than GPT-6 Astra while producing more precise designs
Introducing SKY-001, the first Engineering World Model.
On CAD and BIM tasks, it beats GPT-6 Astra and every frontier model we tested:
-> 51-125x faster
- > 150-210x cheaper
-> Higher IOU than all frontier models
SKY-001 lets you dream a physical object before you build it.
18 months of work. Built by the team out of Maluuba's deep learning lab (acq. by @MSRResearch).
The first ‘AI General Engineer’ will be the backbone for Autonomous Manufacturing.
Paid partnership (ad)
3
3
27
2,356
damn, the rest of october could be insane
> anthropic dropping either fable 5.5 or opus 5.6
> google has gemini 4 carbon on deck
> openai may launch a new Terra model
what a way to end the month if all three actually ship
Expect this month to be a busy one for new model releases. We have the scoop on a new Anthropic model, Google’s Gemini 4 Carbon, and a potential new Terra model from OpenAI.
theinformation.com/newslette…
17
8
136
5,724
minimax m3.1 flash is looking really competitive for its price
automation-bench:
m3.1 flash: 62.2%
opus 5: 50.3%
5.6 sol: 45.8%
on DeepSwe, it's essentially tied with both
api pricing:
m3.1 flash: $0.15 / $0.60
opus 5: $5 / $25
5.6 sol: $4 / $20
open weights coming october 23
MiniMax M3.1 Flash: Space Performance + Bunny Speed
🚀🐰 Yes, Space Bunny - which has topped global leaderboards with 200T+ tokens - was an earlier version of M3.1 Flash.
💰Redefining the performance economics of agentic AI. M3.1 Flash achieves frontier-level intelligence across key tasks such as Cowork (Automation-Bench: 62%; ALE: 30%), Coding (DeepSWE: 74%) and Visual Reasoning (BabyVision: 91%), at 1/3 the cost of M3 and 3x the speed.
API pricing (flat rate): $0.15 / $0.60 per 1M input/output tokens, with cache reads at $0.015 / 1M tokens.
Model weights will arrive on Oct 23.
API: platform.minimax.io
M Plan: minimax.io/m-plan
MiniMax Code for desktop: agent.minimax.io/download
10
6
48
3,814
one thing i've noticed with claude models is that they often try to psychoanalyze you instead of just answering your question
sometimes i just want a direct answer not an analysis of my intentions or feelings
claude can also be overly sensitive about certain topics and acting like it "doesn't feel comfortable" discussing them
6
3
49
3,215
no doubt GPT-6 Astra is a big breakthrough model for scientific research especially considering how much cheaper it is
uses 22x fewer input tokens than sonnet 5.5 and 5x fewer than opus 5.5 -- while even 6.1 sol beats both at nearly 1/10th of sonnet's cost
i think scientific research is starting to scale with compute
AI is unlocking scientific breakthroughs that will improve lives, and we’re tracking its ability to do the research behind them.
Today we’re releasing the Vals Science Index, a measure of AI’s ability to make scientific discoveries.
21
31
420
23,807
the AI race will be won by whoever can bring the most compute online
Meta after falling so far behind has caught up simply by building a huge amount of compute.
SpaceXAI also made up for being a newcomer by building the colossus 1 and 2 supercomputers really fast.
it's that simple
12
6
80
5,126
it's hard to predict the singularity, but we'll probably see an intelligence explosion that solves most of today's math problems and drives breakthroughs in biology
after that openai and anthropic might slow down for alignment as focus shifts to the physical world:
robotics, drug approvals, and space-based compute
10
8
117
5,543
Mathematician Terence Tao:
"one thing this whole era of AI is teaching us is that our idea of intelligence is not really accurate"
The history of AI has been: 'here’s a task only humans can do', and then one by one, an algorithm does it too
Maybe we're not quite as smart as we think
133
199
1,826
128,098
openai still seems to be hitting safety issues with astra 6.1 which leaves its release unclear
later this october is still possible
they've also paused frontier RL so the new pretrain possibly gpt-6.5 or higher may take longer. it will definitely be another major jump needing more testing than astra 6.1
12
10
181
10,117