Matic's cleaning robot sees your home with 5 cameras instead of lidar
Tesla's 2019 vision-only bet has migrated to home robotics — Karpathy's argument, and lidar prices that have fallen to $200
10
AI
AI models, services and robotics
Matic cleaning robot now goes where you point
An update combining voice and gesture matches arm direction to a 3D map to clean a specific spot
60
Gemini 3.7 Flash powers Gemini Spark, boosting Workspace manipulation
Rollout hits Pro and Ultra subscribers in 160 countries, jumping from 3.6 to 3.7 in a single day
10
ChatGPT adds 'Computer History' feature to remember computer usage
Rolled out as opt-in feature in Mac desktop app, follow-up to Chronicle preview
150
Google launches Gemini 3.7 Flash for coding, beats Claude on benchmark
New Flash model spotted a day before launch, priced at half the previous version
40
GPT-5.6 Sol Ultrafast Mode: Up to 14x Faster
OpenAI previews GPT-5.6 Sol's Ultrafast mode, first for select enterprise API customers. A warehouse-layout demo suggests speed without quality loss
190
DeepSeek-V4-Pro launches officially with three-tier reasoning intensity control
Released via Expert Mode, V4-Pro edges out Opus-4.8 on Terminal Bench
40
Suno Studio 2.0 adds MIDI, but still no VST support
Generative music AI Suno moves from composition tool toward editing tool, adding MIDI — its most-requested feature
550
Gemini 3.7 Flash spotted as version rollout accelerates
The name of Google Gemini's next experimental version was found in a filter search box
00
GeForce NOW launches full Linux release, targets Chromebooks
The Linux native app exits beta, and new Chromebook buyers get a year of free access
60
Zoom Screen-Sharing Flaw Cracked With Fewer Than 20 AI Prompts
Vulnerability let attackers remotely control cameras and microphones by abusing the annotation feature, discovered and patched after A Security's find
120
Mistral sells regional server processing but leaves agents out
Regional endpoints and a paid priority tier arrive — only function calls get regional guarantees, files and agents are exceptions
20
NVIDIA's Nemotron 4 aims for 1 trillion parameters, still trails China
NVIDIA's open-weight Nemotron 4 is said to reach 1 trillion parameters, but Kimi K3 and DeepSeek V4 are already bigger
60
Pixel Watch 5 price rises $50 with offline Gemini
Offline-capable Gemini and insulin resistance tracking arrive as price climbs to $399
50
Grok Bot Early User Reviews: The First 48 Hours
74 pieces of game art in two hours, a reservation booked by voice from a parking lot, a file that wouldn't send, and Musk's promise to fix the basics first
3090
'Chief of Staff' bot emerges, coordinating multiple Grok bots to prep a Pokémon stream
A screenshot shared by Wes Roth shows a Grok bot directing coach, host, and player bots
30
Cursor Privately Testing Codebase and Review Tabs
New "Origin" project with GitHub repo sync undergoing internal testing with partners
20
Sakana AI adds new models Namazu and Fugu to its chatbot
Vibe coding you can trigger with words, now expanded into Japanese-specialized models
20
Upstage Solar Pro 4 Intelligence Index jumps from 14 to 42
Score triples over predecessor in Artificial Analysis evaluation, real-world task Elo rises from 498 to 1276, surpassing human baseline
30
Grok 4.6 unveiled: top-tier performance at half the price
Scoring 61 on the intelligence index, it ties GPT-5.6 Sol and closes to within two points of Claude Opus 5 — yet costs less than half as much. A leap achieved not by scaling up but by changing the training recipe alone. The price war sparked by xAI has only just begun.
40
Claude in Chrome adds cross-device session continuity
Skills and connectors used in the browser now carry over to desktop and mobile, starting with Max and Team plans
70
DeepSeek V4 Pro GA Benchmarks Leak, Nears Top Open-Source Tier
Matches Kimi-K3 and GLM-5.2 in agentic evaluations like Terminal Bench and HLE
20
Grok 4.6 pricing revealed — DeepSeek 4 Pro undercuts it further
$2 per million input tokens, $6 for output — cheaper than GPT-5.6 and Claude, but DeepSeek is cheaper still
80
Musk Teases Grok 4.7 Will "Surpass Every Existing Model"
Musk cites SpaceX's unique training data, expressing confidence in the next model
20