ChatGPT vs. Gemini: Daily Active Users

Source : https://www.reuters.com/world/india/with-freebies-openai-google-vie-indian-users-training-data-2025-12-17/... và 115 kí tự

Progress of all Frontier released models from January 1st 2025 till now

Gemini 3 Flash is the most cost-efficient frontier model

Artificial Analysis Intelligence Index score and cost wise.... và 59 kí tự

30 min codex-cli with GPT-5.2 hihg made fully working NES emulator in pure c with mapper 0.

This is why multimodal LLMs won’t stay in browsers

Tencent announces HY-World 1.5. An open source interactive world model that runs at 480p 24 FPS on consumer hardware.

They have it up on their website at https://3d.hunyuan.tencent.com/sceneTo3D. It is not in English and currently has a waiting list. However they have provided the files needed to run it on your own h... và 749 kí tự

Gemini 3 Flash scores 92.0 on the Extended NYT Connections benchmark (Gemini 2.5 Flash scored 25.2, and Gemini 3 Pro scored 96.8).

More info: [https://github.com/lechmazur/nyt-connections/](https://github.com/lechmazur/nyt-connections/)... và 105 kí tự

Reuters: Daily Active Users for AI tools continue to skyrocket, especially in India

Reuters dropped a really interesting piece today on how OpenAI and Google are aggressively rolling out free AI tools in India. The reason is simple and this chart makes it obvious. Massive user growt... và 853 kí tự

Reuters is reporting that China's classified EUV project has reverse engineered and successfully built a prototype EUV machine with the help of former ASML engineers

Chinese EUV Lithography Machine Prototype Reportedly Undergoing Testing

Reuters China completed working EUV machine early 2025

Not Gemini Flash beating Pro on ARC-AGI-2

UPDATE: Independent Benchmarks for Gemini 3 Flash (Highest 'Omniscience' Score ever recorded) + Google Lead teases: 'The week is not over yet.' Gemma 4 incoming?

**Mods: This is a follow-up analysis. This post contains independent datas just now released from Artificial Analysis and new developer comments that were not available in the initial launch post** T... và 1,438 kí tự

Gemini 3.0 Flash beats 3 Pro in SWE Agentic coding

Gemini-3-Flash Artificial Analysis benchmark results.

Impressive results. GPT-5.2 xHigh is not available in web with a 20$ subscription. But Gemini-3-Pro and Flash are accessible for free in aistudio. However, it has a higher hallucination rate than Pro... và 200 kí tự

Google releases Gemini 3 Flash: Ranks #3 on LMArena (above Opus 4.5), scores 99.7% on AIME and costs $0.50/1M plus Benchmarks.

**Sources:** * [Google Blog](https://blog.google/products/gemini/gemini-3-flash/) * [LMSYS Leaderboard Tweet](https://x.com/arena/status/173641...) ... và 152 kí tự

Gemini 3.0 Flash is out and it literally trades blows with 3.0 Pro!

SPOTTED: Gemini 3 Flash is LIVE on Vertex AI & might be rolling out on 2 variants (Fast & Thinking)

**Gemini 3 Flash (preview):** Google's latest workhorse model with enhanced multimodal and coding capabilities, optimised for complex data processing and agentic tasks. **Knowledge cut-off:** Jan 202... và 763 kí tự

Alr Gemini-3-flash is here!

just tested it out and it's amazing! The hype was real. I tested it on a simple website creation prompt and the results are actually good! Gemini-3-flash: https://g.co/gemini/share/df8444809d15 Gemi... và 254 kí tự

GPT Image 1.5 vs Nano Banana Pro realism test