reasoning models already know when they've solved the problem. we just don't let them stop. new paper from Beihang University and ByteDance shows that the overthinking problem in models like DeepSeek-R1 and Qwen3 isn't a training failure. it's a sampling failure. the fix cuts
GENERATIVE AI
-
Anthropic Revenue Growth, Contract Compromise and Nvidia Results
By
–
3-4-2026 ($) •Anthropic's Skyrocketing Revenue
•A Contract Compromise?
•Nvidia Earnings https://stratechery.com/2026/anthropics-skyrocketing-revenue-a-contract-compromise-nvidia-earnings/ [Translated from EN to English]→ View original post on X — @benthompson, 2026-03-04 11:00 UTC
-
AI-Generated Games Adapt Dynamically to User Micro-Engagement
By
–
Future games will be made “for you” on the fly, generating gameplay and content based on your micro-engagement.
-
Question about Gemini Diffusion model progress and speed
By
–
Slight off topic question Demis, but how is the work on the Gemini Diffusion model? If I got it right that model was churning out at ~1500t/s ? Are you (GDM) actively working on this project still?
-
Google Launches Gemini 3.1 Flash-Lite: Fast and Cost-Efficient
By
–
small but mighty 💪 – our new Gemini 3.1 Flash-Lite model is incredibly fast and cost-efficient for its performance https://t.co/rD33f5qTiV
— Demis Hassabis (@demishassabis) 4 mars 2026small but mighty – our new Gemini 3.1 Flash-Lite model is incredibly fast and cost-efficient for its performance
-
Critic mocks OpenAI’s cringe announcement of GPT-5.3 Instant
By
–
Less cringe 🤦. Worst way to announce features of a new model. OpenAI (@OpenAI) GPT-5.3 Instant in ChatGPT is now rolling out to everyone. More accurate, less cringe. openai.com/index/gpt-5-3-ins… — https://nitter.net/OpenAI/status/2028893701427302559#m
→ View original post on X — @skathirmani, 2026-03-04 03:45 UTC
-
Claude AI App Downloads Surge 484 Percent in Two Weeks
By
–
Some astounding stats: App downloads for @claudeai are up 484% in less than two weeks.
-
GPT-5.3-Codex-Spark Available on Poe Platform
By
–
Try it alongside every top model in the Poe app and API today at https://
poe.com/GPT-5.3-Codex-
Spark
…. Pricing is 10 points/message to ensure broad access. -
GPT-5.3-Codex-Spark now available for fast code generation
By
–
GPT-5.3-Codex-Spark is now available on Poe in a limited research preview! Poe is one of the first places where this model is available today. Enabled by Cerebras’s high-performance chips, OpenAI's newest lightweight coding model is optimized for ultra-fast code generation
-
Qwen AI dominance and Alibaba leadership shakeup
By
–
Qwen team:
*each new model family obliterates SOTA*
*bonus vision*
*dense models are ngmi? Not on my watch*
*literally own the Pareto frontier for years*
*ultra based capybara mascot* Alibaba:
*nuke leadership*