#Business
Curated coverage on «Business»: releases, research, benchmarks, open-source.
NEWTG: @ai_machinelearning_big_data (Machinelearning)Pika Labs releases open video model H3 Flex: 6 steps instead of 50
TG: @ai_machinelearning_big_data (Machinelearning)OpenAI Rewrites GPT Behavior Rules: Five Values and a Chain of Command
Habr AIVoice bots built on LLMs turn to Jev, a fast classifier, for decisions in 200–500 ms
Latent Space (swyx)Google admits AI Mode cuts search revenue, shares fall 15%
arXiv cs.AIThinkJev Makes AI Think Only About Hard Questions: 28% Fewer Tokens
Habr AIAI cartoon series: a 24-second episode built ten ways, the best cost 162 credits
AWS Machine Learning BlogWorldwide Express hands disputed invoices to AI agents and cuts unresolved disputes by 75%
Habr AIOpen-source LLM gateway KTMS 1.0 plugs your own GPUs into the cloud and survives outages
ThinkJev Makes AI Think Only About Hard Questions: 28% Fewer Tokens
One Qwen3.5 model answers instantly when it is confident and switches on reasoning when it is not. Across 8359 decisions, it uses 28% fewer tokens while accuracy…
AI cartoon series: a 24-second episode built ten ways, the best cost 162 credits
The makers of "Forest Street" spent three days testing Wan, Kling, Veo, Seedance and GPT Image. A second-by-second script and a cheap image-based draft won…
Worldwide Express hands disputed invoices to AI agents and cuts unresolved disputes by 75%
The logistics giant deployed three AI agents on Amazon Bedrock. They close about 10,000 disputed carrier invoices a week and have freed up $10 million in frozen…
Google: AI Agents Using Android CLI Complete UI Tasks 77% Faster
VK cuts the launch of a new recommendation feed from weeks to a single day
Claude Enterprise Gets Smart Reports: Who Uses AI How, and What It Costs
Claude Opus 5.5 scores 68% on HLE-Diamond, but each task costs 35 times more
Google's Gemini 4 Carbon Scores 88.2% on Vulnerability-Finding Benchmark, Access Remains Restricted
Google DeepMind unveiled Gemini 4 Carbon, which scores 88.2% on SEC-bench Pro, a test of finding real software vulnerabilities. For now, only some cyber defenders…
TokenRouter speeds up token-level model routing by up to 64x
Researchers built a runtime that switches generation between a small and a large LLM at every token, raising throughput 2–64x over earlier implementations.
Mistral Releases Large 4 Weights: Open 1T-Parameter Model for Coding
Anthropic will give companies token reports: customer spending grew 9x in a year
ElevenLabs hands routine work to internal agents and saves 1,200 hours a month
GitLab Orbit hits GA: agents need 4.5x fewer tokens
Gemini 4 Carbon tops independent rankings, but only cyber defenders can use it
Google DeepMind released Gemini 4 Carbon: 59 points on the Artificial Analysis index vs. 58 for Opus 5.5. For now, access is limited to vetted cybersecurity…
How OpenAI Kept Azure Cosmos DB Standing Under the Weight of 1 Billion ChatGPT Users
OpenAI engineers explain how they rebuilt their Azure Cosmos DB storage architecture three times — from a shared database to separate containers and back again — to…