AI models

News and updates on leading AI models - from ChatGPT and Claude to Gemini and open-source innovations.

Usman Salis Usman Salis AI models
6 months ago

Cursor's Composer 2 Hits 61.3 on CursorBench With $0.50/M Token Pricing

Cursor's Composer 2 hits 61.3 on CursorBench at $0.50/M tokens, challenging top AI coding models on price-performance.

Peter Smith Peter Smith AI models
6 months ago

Why Multi-Agent AI Systems With Specialized Roles Are Outperforming Single Models

Specialized multi-agent AI systems are replacing single models, boosting efficiency through coordinated roles and shared memory layers.

Eseandre Mordi Eseandre Mordi AI models
6 months ago

MiniMax M2.7 Hits 56.2% on SWE Bench, Challenging Top AI Models

MiniMax M2.7 debuts with self-evolving training, 30% gains, and benchmark scores rivaling Claude and GPT-5

Peter Smith Peter Smith AI models
6 months ago

MiniMax M2.7 Hits 66.6% Medal Rate With Self-Evolving AI Across 22 Competitions

MiniMax M2.7 self-improves across 22 ML competitions, hitting 66.6% medal rate with no human input.

Saad Ullah Saad Ullah AI models
6 months ago

Google Open-Sources Colab MCP Server, Giving AI Agents Cloud Code Execution

Google open-sources Colab MCP Server, letting AI agents run cloud code via Claude, Cursor, Codex, and Gemini.

Marina Lyubimova Marina Lyubimova AI models
6 months ago

Google and OpenAI Lead AI Image Race With 60+ Models Ranked

Google and OpenAI dominate Design Arena AI image benchmarks. Gemini 3.1 Flash and ChatGPT models top rankings.

Usman Salis Usman Salis AI models
6 months ago

MiniMax M2.7 Launch: 200K Context and 56% SWE-Pro Score Enter the Agent Race

MiniMax M2.7 launches with 200K context and 56.22% SWE-Pro score, targeting AI agent workflows.

Usman Salis Usman Salis AI models
6 months ago

GPT-5.4 Pro Solves 2 of 100 Unsolved Math Problems in Oxford Benchmark

GPT-5.4 Pro cracked 2 unsolved math problems in Oxford's benchmark, outperforming rivals with a 7% solve rate.

Saad Ullah Saad Ullah AI models
6 months ago

Cursor AI Cuts Token Use 5x With Self-Summarization RL

Cursor's RL self-summarization beats compaction, scoring 47.9 on CursorBench Hard 80k with 5x fewer tokens.

Usman Salis Usman Salis AI models
6 months ago

Grok Imagine Claims Top 3 DesignArena Video Benchmarks with 1337 Elo

xAI's Grok Imagine video model ranked first in Video Arena, Image-to-Video Arena, and Video Editing Arena on DesignArena leaderboards.

Eseandre Mordi Eseandre Mordi AI models
6 months ago

Gemini 3 Pro Hits 82.2% Accuracy, Matching Humans on MADQA Benchmark

A new MADQA study shows Gemini 3 Pro matching human accuracy at 82.2%, but humans and AI still solve different problems in completely different ways.

Alex Dudov Alex Dudov AI models
6 months ago

Grok Growth Surges as 42.8% of Grok.com Visitors Are Brand-New Users

xAI's Grok attracts 42.8% new users while retaining loyal returning visitors across its growing AI platform.

Peter Smith Peter Smith AI models
6 months ago

ChatGPT Captures 82.86% Returning Users, Leading All Gen AI Platforms in February 2026

Similarweb's February 2026 data reveals ChatGPT dominates user retention across generative AI platforms, with rivals like Gemini and Perplexity showing solid but smaller returning visitor shares.

Eseandre Mordi Eseandre Mordi AI models
6 months ago

Grokipedia v0.2 Hits 500,000 Edits Challenging Wikipedia With AI Knowledge

xAI's AI-generated encyclopedia reaches 500K approved edits, positioning itself as a new alternative to Wikipedia.

Saad Ullah Saad Ullah AI models
6 months ago

IBM's Agent Memory System Boosts Task Success by 149%

IBM's new AI framework lets agents learn from past executions, lifting complex task completion from 19% to 47% without retraining.

Alex Dudov Alex Dudov AI models
6 months ago

Grok 4.20 Hits 96.5% Accuracy and Ranks #2 in Telecom Benchmark

xAI's Grok 4.20 scores 96.5% on the tau2-Bench Telecom test, outpacing Claude Opus 4.6, GPT-5.4, and Gemini 3.1 Pro.

Saad Ullah Saad Ullah AI models
6 months ago

Grok 4.20-Beta1 Claims #2 Spot on Search Arena With Score of 1225

xAI's Grok 4.20-beta1 lands second on Search Arena, putting it among the most competitive AI search systems right now.

Usman Salis Usman Salis AI models
6 months ago

Cursor Benchmarks GPT-5.4 Coding Efficiency: 60%+ Score on CursorBench

Cursor launched CursorBench to score AI models on agentic coding tasks, measuring both intelligence and token efficiency across 10+ leading models.

Usman Salis Usman Salis AI models
6 months ago

NVIDIA's AutoGaze Cuts 99% of Video Tokens, Speeds Up AI Models 19x

Researchers from NVIDIA, MIT, UC Berkeley, and Clarifai built AutoGaze - an AI filter that slashes video data by up to 99% without hurting model accuracy.

Saad Ullah Saad Ullah AI models
6 months ago

Lightpanda Browser Claims 11x Faster AI Automation with 9x Less Memory

Lightpanda open-source headless browser runs 11x faster than Chrome with 9x less memory for AI agents.

Posts feed
Yandex Releases Open-Source Alice AI Foundation 80B Model
2 weeks ago
Xiaomi Releases MiMo-V2.6 Open-Source AI Models
2 weeks ago
Early Access to Fable 5.2 Shows Strong Performance
2 weeks ago
Anthropic Prepares to Launch Opus 5.5 AI Model
2 weeks ago
NVIDIA BlueField-4 Expands DPU Control From 400 Gb/s to 7.2 Tb/s
3 weeks ago
Alphabet Gets a $485 Price Target as Tigress Bets on Its Full-Stack AI Strategy
3 weeks ago
Google Launches Gemini 3.8 Live and Extended Thinking
3 weeks ago
Meta One Unifies Paid Features Across Instagram, Facebook, WhatsApp and Meta AI
3 weeks ago
NVIDIA Expands CUDA-Q for Fault-Tolerant Quantum Computing
3 weeks ago
Anthropic Released a Video Guide on Loop Engineering
3 weeks ago
Follow us on Twitter

Aigazine is an independent publication covering artificial intelligence — new models, research, the companies building them, and the industries they reshape. Published from Tbilisi, Georgia (0179, Tbilisi City, Vake District, 49 Besarion Zhghenti Street, VAT 305786600).

© 2026 Aigazine

We use cookies to improve your experience on our site and to show you relevant advertising. To find our more, read our privacy policy and cookie policy