Sambhav Kothari
@_sambhavkothari
Followers
407
Following
6K
Media
91
Statuses
3K
Head of AI Productivity @bloomberg. Opinions are my own. https://t.co/2B0HQdJeqq
London, England
Joined November 2016
- no benchmarks - no API release - system card looks sloppy at best and is weirdly short - was planned for late nov maybe their weirdest new model launch ever? why did they randomly rush it out..?
What the fuck lmao they dropped GPT-5.1 out of nowhere with no benchmarks in sight Are they trying to beat someone to the punch
54
23
920
GPT-5.1 is more efficient and better at deciding how much time to spend on reasoning
20
13
206
OpenAI is now the Apple of AI and there’s no going back. When people say “AI,” they don’t mean Anthropic, Gemini, or Grok, they mean ChatGPT. Just like when people say “tablet,” they mean iPad. When they say “smartphone,” they mean iPhone. OpenAI didn’t just build a model, they
474
105
929
Pretty amazing: DeepMind just taught AI to see more like humans. By aligning visual models with how people actually group and interpret images, they make perception systems more intuitive, robust, and generalizable. It’s a step toward AI that understands meaning, not just
Humans think about things conceptually – like how cats and starfish are both animals, despite their differences. But AI can sometimes miss this nuance. 🖼️ Our research teaches vision models to better organize visual concepts, making them more reliable and better at generalizing.
19
55
635
Also, we've made it easier to customize ChatGPT. You can pick from presets (Default, Friendly, Efficient, Professional, Candid, or Quirky) or tune it yourself.
428
212
4K
GPT-5.1 is out! It's a nice upgrade. I particularly like the improvements in instruction following, and the adaptive thinking. The intelligence and style improvements are good too.
2K
1K
14K
I've been testing GPT-5.1 for a few days. My quick notes: - creative writing style is a LOT better - it's much faster than GPT-5 (with similar intelligence) for most prompts - the personality is WAY better (but can still sometimes be annoying) - it's great in Codex!
GPT-5.1 is out! It's a nice upgrade. I particularly like the improvements in instruction following, and the adaptive thinking. The intelligence and style improvements are good too.
63
42
955
For the first time, Anthropic is building its own AI infrastructure. We’re constructing data centers in Texas and New York that will create thousands of American jobs. This is a $50 billion investment in America. https://t.co/hmJXNODQsG
anthropic.com
Anthropic is an AI safety and research company that's working to build reliable, interpretable, and steerable AI systems.
210
414
4K
We built a Deep Research demo for the Claude Agent SDK! It's one our most requested use cases: spawn multiple AI agents to research a topic in parallel, then synthesize their findings into a report. 🧵 on how it works:
26
49
713
Huge: Yann LeCun is about to leave Meta! He plans to launch his own start-up! That was to be expected. Resources were increasingly taken away from him, and important positions were reassigned. At the same time, Lecun is also largely responsible for Meta's significant losses in
105
78
1K
agents need state, storage and streaming durable objects have a SQLite db, s3 api and native websocket support
sandboxes is the next major infrastructure primitive there’s going to be a couple more too all infra will need to be rethinked with the assumption that state and maintaining that state is a requirement
9
5
166
sandboxes is the next major infrastructure primitive there’s going to be a couple more too all infra will need to be rethinked with the assumption that state and maintaining that state is a requirement
17
2
99
MCP is for networked apps, not libraries. Most of the cognitive dissonance with MCP comes from using it as if it were a library - its original intention. That said, it’s evolved and is evolving very quickly away from its origins as it found a better, more useful use case:
I’m still fairly bullish on mcp. I think the big unlock will be when people start loading background agents with them. I realize this is possible rn, but a background agents are still nascent. The local agent with mcp server paradigm does not make sense. The primary benefit of
1
2
9
when we shipped code mode in cloudflare workers a couple months ago, it came from intuiting that LLMs were better at writing code than tool calling (since they were trained on lots of code!) the fact that it can reduce token usage (read: costs) by 99% is 🤯
5
18
228
At Context7, we benchmark LLMs for how well they enrich context — not just accuracy. 10 = best contextual quality 0 = worst Here are the latest results 👇
39
27
478
I don't know what I'm more excited about: Gemini 3.0 or GPT-5.1. Rumors are Gemini 3 reveal/release November 12th and GPT-5.1 November 24th. So the next few weeks could be really exciting.
72
24
818
BREAKING 🚨: OpenAI is preparing to ship GPT-5.1 family with 3 new models: - GPT-5.1 - GPT-5.1 Reasoning - GPT-5.1 Pro These traces appear in the code for Enterprise RBAC settings, and the release date is set to "November 24". What if... 👀
BIG OPENAI NEWS: GPT-5.1, GPT-5.1 Reasoning and GPT-5.1 Pro GPT-5.1: "Flagship model for the latest generation of ChatGPT" GPT-5.1 Reasoning: "Thinks longer for better answers" GPT-5.1: "Research-grade intelligence"
43
85
871
After two years of work, we’ve made an AI Scientist that runs for days and makes genuine discoveries. Working with external collaborators, we report seven externally validated discoveries across multiple fields. It is available right now for anyone to use. 1/5
111
548
4K