This week in AI infra
Every deprecation, API change, pricing shift, and model update that could affect builders shipping on AI APIs -- risk-classified and ready to share.
High-impact changes
30 changes across 5 providers.
MiniMax-Music3 momentum +30%
MiniMaxAI model showing momentum in AI Model.
Action: Run model migration checks for quality, latency, and cost.
Muse-Glimmer-30B momentum +4%
meta-models model showing momentum in AI Model.
Action: Run model migration checks for quality, latency, and cost.
MiniMax-H3 momentum +1%
MiniMaxAI model showing momentum in AI Model.
Action: Run model migration checks for quality, latency, and cost.
For more on how Claude ran this experiment and the full results, see our blog: https://t.co/48COrpUcJn
For more on how Claude ran this experiment and the full results, see our blog: https://t.co/48COrpUcJn
Designing a binder is an easier process than designing a drug, but it’s a useful proxy. The typical success rate in the field today is betwe
Designing a binder is an easier process than designing a drug, but it’s a useful proxy. The typical success rate in the field today is between 10% and 15%. Between 22% and 35% of Claude's designs bound successfully, depending on the setup. Some of its strongest designs bound https://t.co/nIVU00ZgiH
DeepSeek V4 Pro, hosted in the U.S., is now available in Perplexity Computer. We evaluated it against other models on WANDR. It scored 0.359
DeepSeek V4 Pro, hosted in the U.S., is now available in Perplexity Computer. We evaluated it against other models on WANDR. It scored 0.359 at $0.75 per task, 62% cheaper than the next model on the cost-performance frontier. https://t.co/uSlRUfcNoI
One prompt → your next landing page. We hooked up Gemini 3.7 Flash with Nano Banana and Omni in @Antigravity to build an interactive landing
One prompt → your next landing page. We hooked up Gemini 3.7 Flash with Nano Banana and Omni in @Antigravity to build an interactive landing page from scratch, generating the copy, images, and video in a single shot. https://t.co/oCyMQN7rvp
You stay in full control of what your agents, and your team’s Claude sessions, can access. Sign in securely using OAuth, with no server to r
You stay in full control of what your agents, and your team’s Claude sessions, can access. Sign in securely using OAuth, with no server to run or API keys to manage. Connect it here: https://t.co/KBThZacFnC https://t.co/Vz9rGs8vkb
Try the @whop connector, now released in Grok https://t.co/9PTh5CWfM6
Try the @whop connector, now released in Grok https://t.co/9PTh5CWfM6
Introducing the ElevenLabs MCP, now available in Claude. Your team can manage your voice and chat agents where you already work - review rec
Introducing the ElevenLabs MCP, now available in Claude. Your team can manage your voice and chat agents where you already work - review recent performance, create new agents, update configurations, and even estimate LLM costs before changes go live. https://t.co/RkGqlr3I0h
SRCC, EY India launch 12-month AI-led programme in machine learning, data analytics, blockchain for students
SRCC and EY India-led programme will follow a selective admissions process, consisting of assessments, interviews, and submission of a Statement of Purpose (SOP).
Action: Benchmark candidate model behavior before adopting in production.
Telecom Cloud Market: Hybrid Cloud Adoption Accelerates
The telecom industry is in the middle of one of its biggest infrastructure overhauls in decades. Operators that once ran everything on dedicated, hardware-bound networks [read full press release...]
Action: Benchmark candidate model behavior before adopting in production.
This AI data center material is facing a price spike
The artificial intelligence (AI) revolution is triggering a spike in demand for optical modules, which facilitate rapid data transfer in data centers.
Action: Recompute cost guardrails and routing thresholds.
Qwen3.8-27B momentum +17%
Qwen model showing momentum in AI Model.
Action: Run model migration checks for quality, latency, and cost.
Qwen3.8-27B-GGUF momentum +1%
unsloth model showing momentum in AI Model.
Action: Run model migration checks for quality, latency, and cost.
Qwen3.8-2.4T-A95B momentum +30%
Qwen model showing momentum in LLM.
Action: Run model migration checks for quality, latency, and cost.
LTX-2.5 momentum +2%
Lightricks model showing momentum in AI Model.
Action: Run model migration checks for quality, latency, and cost.
DeepSeek-V4-Pro-0813 momentum +19%
deepseek-ai model showing momentum in LLM.
Action: Run model migration checks for quality, latency, and cost.
Qwen3.8-27B-FP8 momentum +1%
Qwen model showing momentum in AI Model.
Action: Run model migration checks for quality, latency, and cost.
Qwen3.8-27B-Uncensored-FP8 momentum +12%
orcarouter model showing momentum in AI Model.
Action: Run model migration checks for quality, latency, and cost.
MemoMind One Smart Glasses Launch on Kickstarter For $399
The Smart Glasses Guy explores the MemoMind One, a wearable designed to enhance productivity through practical functionality. A standout feature is its dual high-brightness displays, which deliver up to 2,000 nits of brightness for clear visibility even under direct sunlight. These displays can be customized to separate key information into distinct sections, reducing visual clutter [...] The post MemoMind One Smart Glasses Launch on Kickstarter For $399 appeared first on Geeky Gadgets .
Action: Benchmark candidate model behavior before adopting in production.
ASEAN digital pact offers India model for deeper Southeast Asia ties: Report
A new report says ASEAN’s Digital Economy Framework Agreement offers India a model for deeper digital engagement with Southeast Asia, promoting interoperable digital trade, payments, data flows and emerging technologies.
Action: Benchmark candidate model behavior before adopting in production.
The Download: how people really use AI, and Flock’s design choices
This is today’s edition of The Download, our weekday newsletter that provides a daily dose of what’s going on in the world of technology. We still don’t know how people are really using AI AI companies like Anthropic and OpenAI regularly publish reports on how people are using their products. But they only release the...
Action: Benchmark candidate model behavior before adopting in production.
AI/R Compass UOL Named Amazon Quick SI Partner by AWS Generative AI Innovation Center
AI/R Compass UOL, a subsidiary of AI/R—a technology company specialized in agentic AI—has been named as an Amazon Quick SI Partner by the AWS Generative AI Innovation Center (GenAI IC), enabling AI/R Compass UOL to accelerate the deployment of production-ready agentic AI solutions that deliver measurable business outcomes for enterprise clients. Through the launch of [...] The post AI/R Compass UOL Named Amazon Quick SI Partner by AWS Generative AI Innovation Center appeared first on AiThority .
Action: Benchmark candidate model behavior before adopting in production.
Prodigy Technovations Unveils Industry's First UFS 5.0 Protocol Exerciser and Analyzer to Power the AI-Driven Storage Era
BENGALURU, India, Aug. 18, 2026 /PRNewswire/ -- Prodigy Technovations Pvt. Ltd, a leader in protocol analysis solutions, proudly announces the successful testing and launch of the PGY-UFS5-EX-PA. ...
Action: Benchmark candidate model behavior before adopting in production.