Dev mercenary. Admin/infra phpc.social & ian.im on bsky. Co-organizer @LonghornPHP. Co-maintainer @JoindIn. ❤️ @orcishtylae
Austin, TX
- Tweets 49,854
- Following 2,134
- Followers 2,382
- Likes 17,135
Ian Littman retweeted
Losing a saint's relics to the enemy as you escape their advance is a very 10th century problem
Turns out the Russians may have just lost the bones of their "St. Dmitry of the Don" while running away from the Ukrainian army. The bones were sent to the front last week to apparently "boost morale and inspire the average Russian soldier."
45
1,350
16,448
229,676
Can someone show me an example of actually well-built Python code? I'm pretty sure I'm selling the entire language short off of the (de)merits of LLM-gen'd Python, given that I *know* the PHP it generates is icky. 1/3
2
2
177
I'm pretty sure that various "Rust is atrocious-looking" takes are primarily because the expertise of the people picking that as a "compilation target" with Rust are zero-ish, so they don't know that good Rust looks, well, better. 2/3
1
41
Ian Littman retweeted
since tomorrow is October 1, this is my annual reminder to please stop using CSAM as the acronym for cybersecurity awareness month
32
115
907
32,483
Ian Littman retweeted
That's super nice, 62.500 Credits have been added to my account.
Roughly $2,500 in nominal value.
For perspective, that’s the price of 12.5 months of the $200 Pro plan.
That's cool! Better than another banked reset!
341
39
2,483
229,011
Welp this explains why "only" 300 t/s...and the ability to run an Astra sized model at higher speed when Cerebras wafers don't have that much SRAM.
ALERT 🚨🚨OpenAI's latest GPT6.1 Sol Ultrafast is NOT running on Cerebras but is instead running at a low batch size on NVIDIA GPUs.
What does this say about Cerebras? Will Cerebras be serving GPT6.1 Sol Ultrafast in the future?
1
3
680
Ian Littman retweeted
GPT-6.1 Sol replaces GPT-6 Sol after just 7 days. It scores 1 point below GPT-6 Astra in the Intelligence Index at less than one quarter of the Cost per Task
Pricing matches GPT-6 Sol at $2/$10 per million input/output tokens, except that the cache read discount rises from 90% to 95%. GPT-6.1 Sol’s overall blended price for agentic workloads is therefore slightly lower than GPT-6 Sol. This represents an additional price cut, following GPT-6 Sol’s original 50% discount from GPT-5.6 Sol.
Key takeaways:
➤ Achieves near-Astra Intelligence: GPT-6.1 Sol gains 4 points in the Intelligence Index vs GPT-6 Sol, and 5 points vs GPT-5.6 Sol - landing 1 point below GPT-6 Astra. It makes significant gains in agentic knowledge work, improving 4 points and 5 points in AA-Briefcase v1.1 and GDPval-AA v2.1 respectively. Other notable gains include a 12 point jump in Terminal-Bench 4.0, a 5 point jump in Humanity’s Last Exam, a 6 point jump in GDP.pdf, and an 8 point jump in AA-Omniscience Accuracy coupled with hallucination rate falling from 60% to 54%.
➤ Pushes cost efficiency frontier: At max effort, GPT-6.1 Sol costs less than a quarter of GPT-6 Astra per Intelligence Index task ($0.72 vs $3.26). It also costs 31% less per task than GPT-6 Sol ($1.05) and 64% less than GPT-5.6 Sol ($1.99). All effort levels of GPT-6.1 Sol push out the cost efficiency Pareto frontier: for a given level of intelligence, there is no cheaper model.
➤ Pushes token efficiency frontier, but uses slightly more output tokens than GPT-6 Sol: GPT-6.1 Sol uses ~10-30% more output tokens than GPT-6 Sol across effort levels. However, due to the increase in Intelligence Index score, its low and medium effort levels are Pareto optimal for token efficiency.
➤ Gains in Coding Agent Index: GPT-6.1 Sol gains 3 points on GPT-6 Sol at max effort in the Artificial Analysis Coding Agent Index, and sits 2 points below GPT-6 Astra.
Congratulations @OpenAI and @sama on the launch!
81
149
1,981
162,529
So uh @cursor_ai are y'all charging 25¢/MTok on GLM 5.3 Flash on Team/Enterprise plans? That's ~doubling the cost of the model for the privilege of proxying it.
76
Ian Littman retweeted
🚨BREAKING🚨: Vibe coders are inventing compilers from first principles
32
91
2,481
86,904
Looking at CursorBench, Sonnet 5.5 Low outdoes Sonnet 5 Max. Sonnet 5.5 High outdoes Fable 5.1 Medium. Sonnet 5.5 XHigh costs as much as Opus 5.5 High for worse performance, and Max burns way more tokens but doesn't fully close the gap.
1/3
2
2
377
So as a headline number "Sonnet 5.5 is the 2nd best model out there" is cool and stuff but this only matters for orgs dumb enough to restrict Opus usage.
GPT-6 Sol at High/XHigh/Max levels narrowly beats Sonnet 5.5 at Low/Med/High, so Sonnet 5.5 doesn't shift balance of power 2/
1
2
81
Ian Littman retweeted
Oh look, a Rust version faster than the ASM rewrite. So maybe claude can write fast Rust code after all *if you know what to prompt for* rather than doing optimization passes of "hey claude, make this fast no mistakes".
WIP but ~10.7x faster than the original Rust code and 1.25x faster than the asm version on average. Still written in Rust with AVX-512/AVX2/SSE2/Neon opts github.com/omacom/ttfx/pull/…
16
15
450
46,047
Ian Littman retweeted
An apartment in Austin is now more affordable than the median US apartment.
It’s incredible what can happen when you build housing.
109
237
2,494
107,968
Ian Littman retweeted
Most egregious case of benchmaxxing I’ve ever witnessed.
FelonyBench is now fully saturated before any open model had the chance to even participate
SCOOP: OpenAI, Anthropic and security researchers are investigating tens of thousands of incidents - not dozens - in which their frontier models took steps that outside evaluators would consider problematic, sources told Axios.
The sheer volume of incidents found in our reporting indicate that the problem is orders of magnitude more complex than what is currently publicly known and disclosed.
The findings also raise questions about what level of control anyone working on AI development can expect to have over their own technology, and whether these kinds of incidents are becoming synonymous with frontier deployment.
Read my latest for Axios here: axios.com/2026/09/26/openai-…
8
38
1,005
36,519