Workflow
Matthew Berman
icon
Search documents
Elon VS OpenAI (I read the diary)
Matthew Berman· 2026-01-18 03:04
If you're ever planning on becoming rich and famous, don't write anything down. Elon was one of the first backers of OpenAI. At that time, it was a nonprofit, meaning he donated those funds.For years, they were just an AI lab. They had no products, no commercialization of those products. Then in about 2017, after a disagreement about the direction of the company, Elon Musk wanted to become CEO, ended up just leaving.Fast forward a couple more years and OpenAI, Sam Alman, Greg Brockman realized how much mone ...
MIT Researchers DESTROY the Context Window Limit
Matthew Berman· 2026-01-17 19:44
MIT basically solved unlimited context windows and you can apply this to any model. This is called recursive language models and it's just another example of how scaffolding building out infrastructure around the core intelligence of the model still has so much room to grow. Let me tell you about this paper because it is kind of incredible.We study allowing large language models to process arbitrarily long prompts through the lens of inference time scaling. And let me just show you the highle results and th ...
Forward Future Live | 01.16.26 | Guests from Gauntlet AI, Box, Harmonic, and Ramp
Matthew Berman· 2026-01-17 08:18
[music] Heat. Heat. Hey, hey, hey. >> [music] >> Hey, hey, hey.Heat. Heat. Hey, hey, hey, hey, hey, hey, hey.>> [music] >> All right. Welcome to Forward Future Live. It is Friday, January 16th.I'm Matt Berman. >> And I'm Nick Wentz. >> And today we have another incredible show for you.four really incredible guests. We have Austin Allred, founder of Gauntlet AI. Aaron Levy, welcoming him back to the show, co-founder and CEO of Box. Tudor Akim, co-founder and CEO of Harmonic, you've probably heard of him or h ...
New Speedreading Method
Matthew Berman· 2026-01-16 17:47
Maybe I need to make my own tool that does this Follow me ► x.com/matthewberman Instagram ► https://www.instagram.com/matthewberman_ai Media/Sponsorship Inquiries ► https://bit.ly/44TC45V ...
Hired my computer btw
Matthew Berman· 2026-01-15 22:22
Your computer is going to be able to work for you 24/7. This is the future of productivity. Anthropic just dropped Claude Co-work and it is basically Claude Code but for everybody.And if you're not familiar with Claude Code, it is the coding agent that all engineers are talking about right now. So what Enthropic did was take the power of Cloud Code, wrap it in a really nice UI, and then give it to non-developers. Now you can tell your computer to do anything for you.Create presentations, clean up files, sav ...
ChatGPT but 100x Faster
Matthew Berman· 2026-01-15 18:24
chat GPT but 100 times faster. Open AAI just announced that they are partnering with Cerebrris on a three-year multi-billion dollar deal to deliver the fastest inference possible. This deal is worth more than $10 billion and they're agreeing to purchase $750 megawatt of compute power over 3 years.So, a relatively short amount of time, but I can clearly see a future in which OpenAI acquires Cerebrris. And interestingly enough, we just had Andrew Feldman, the CEO of Cerebrris on our live show, but he didn't s ...
Forward Future Live | 12.19.25 | Guests from Google, Deepmind, SemiAnalysis, and Robinhood
Matthew Berman· 2025-12-19 17:57
Guest Lineup: Doug O'Laughlin (President, SemiAnalysis), Koray Kavukcuoglu (Chief AI Architect, Google), Logan Kilpatrick (Group Product Manager, Google Deep Mind), Abhishek Fatehpuria (Vice President of Product, Robinhood) Download Humanities Last Prompt Engineering Guide (free) 👇🏼 https://bit.ly/4kFhajz Download The Matthew Berman Vibe Coding Playbook (free) 👇🏼 https://bit.ly/3I2J0YQ Join My Newsletter for Regular AI Updates 👇🏼 https://forwardfuture.ai Discover The Best AI Tools👇🏼 https://tools.forwardfut ...
The Big Lie About Smarter AI Models
Matthew Berman· 2025-12-19 17:28
Model Improvement & Usage - Model improvements to core intelligence don't significantly impact consumer-focused chatbot users [8][12] - For many personal use cases, speed is more important than the absolute best response, as long as the response is correct [7][13] - Reasoning models are not as crucial for Chat GPT, whose users often prioritize fast answers [14] - Consumers may not fully understand the range of topics the chatbot can address, even with faster, non-reasoning models [17] OpenAI's Internal Dynamics - Tension exists between the research and product divisions within OpenAI [20] - OpenAI had to shift compute resources from research to product development to meet demand, sacrificing future advancements for present needs [27][28] Market Perception & Competition - The perception exists that Anthropic is geared towards enterprise users, while OpenAI is more consumer-focused [9][10] - OpenAI's Chat GPT has strong brand recognition among consumers, which helps it gain traction in the enterprise market [32][33][34] - Integrations and distribution are crucial for success, representing a significant challenge for OpenAI [15][35][36]
Forward Future Live | 12.19.25 | Guests from Google, Deepmind, SemiAnalysis, and Robinhood
Matthew Berman· 2025-12-19 17:27
Guest Lineup: Doug O'Laughlin (President, SemiAnalysis), Koray Kavukcuoglu (Chief AI Architect, Google), Logan Kilpatrick (Group Product Manager, Google Deep Mind), Abhishek Fatehpuria (Vice President of Product, Robinhood) Download Humanities Last Prompt Engineering Guide (free) 👇🏼 https://bit.ly/4kFhajz Download The Matthew Berman Vibe Coding Playbook (free) 👇🏼 https://bit.ly/3I2J0YQ Join My Newsletter for Regular AI Updates 👇🏼 https://forwardfuture.ai Discover The Best AI Tools👇🏼 https://tools.forwardfut ...
AI News: Gemini 3 Flash, GPT Image 1.5, NVIDIA Nemotron 3, Bernie Sanders DOOMER, and more!
Matthew Berman· 2025-12-18 20:52
Gemini 3 Flash is here and it is extremely fast, nearly as good as Gemini 3 Pro and on some benchmarks, even better. It's super efficient and very cheap. It is the best overall model on the planet right now.Here are the benchmarks. So, first the price. It's 50 cents per million input tokens, which is a fourth of the cost of Gemini 3 Pro, a sixth of the cost of Claude Sonnet 4.5%, and a third of the cost of GPT 5.2%. And as you can see here, it is comparable on almost every single major benchmark against Gem ...