In partnership with
Today: xAI puts Grok 4.3 on Amazon Bedrock, OpenAI opens a limited preview of GPT-5.6 and retires GPT-4.5, and Qualcomm steps into data-center silicon with Meta and Microsoft. Here are six developments worth knowing from the past few days.
A WORD FROM OUR SPONSOR
The best prompt engineers aren't typing. They're talking.
Power users figured this out early: speaking a prompt gives you 10x more context in half the time. You include the edge cases, the examples, the tone you want — because talking is fast enough that you don't skip them.
Wispr Flow captures everything you say and turns it into clean, structured text for any AI tool. Speak messy. Get polished input. Paste into ChatGPT, Claude, Cursor, or wherever you work.
89% of messages sent with zero edits. 4x faster than typing. Works system-wide on Mac, Windows, and iPhone.
Models & Products
xAI brings Grok 4.3 to Amazon Bedrock. xAI's Grok 4.3 is now generally available through Amazon Bedrock, where the company lists a 1-million-token context window and configurable reasoning levels. xAI says the model posts the lowest hallucination rate among the frontier systems it benchmarks. The company also shipped Grok Imagine Video 1.5 and a free Grok add-in for Microsoft Word that drafts and rewrites documents using web and X data. (xAI) (Releasebot)
OpenAI opens a limited preview of GPT-5.6. OpenAI began a limited preview of GPT-5.6, a tiered lineup it calls Sol, Terra, and Luna, citing new reasoning modes and stronger coding, science, and cybersecurity performance. Access is restricted to a small group of trusted partners for now, with wider availability planned in the coming weeks. Under its Preparedness Framework, OpenAI rated the models “High” capability in both cybersecurity and biological and chemical risk—below the “High” threshold for AI self-improvement—and said it added tailored safeguards for each. The preview follows June 26 reports of a staggered, government-coordinated rollout. (MarkTechPost) (OpenAI)
OpenAI retires GPT-4.5 and o3 from ChatGPT. OpenAI removed GPT-4.5 and o3 from ChatGPT after a 30-day sunset, with the change taking effect June 27. Conversations that used GPT-4.5 now continue on GPT-5.5, and the company described the move as consolidating its lineup around newer models. Both models remain available through the API, so existing developer integrations are unaffected. (OpenAI) (AI Weekly)
Industry & Business
Qualcomm unveils Dragonfly data-center chips, with Meta and Microsoft on board. At its June 24 investor day, Qualcomm introduced Dragonfly, a data-center silicon portfolio that includes the C1000 server CPU and the AI300 inference accelerator, and named Meta and Microsoft as anchor customers. Meta signed a multi-generation agreement to deploy the C1000 CPU, which Qualcomm said ships in the second half of 2028. The company also confirmed its roughly $3.92 billion acquisition of AI software firm Modular and set a $15 billion data-center sales target for fiscal 2029. (CNBC) (DataCenterDynamics)
Policy & Security
U.S. clears Anthropic's Mythos 5 for about 100 trusted partners. The U.S. government granted Anthropic permission to release its Mythos 5 model to roughly 100 vetted companies and federal agencies, according to a June 26 report. Commerce Secretary Howard Lutnick wrote that appropriate safeguards were in place to let “certain trusted partners” access the model. The decision follows an earlier directive that had barred foreign nationals from Anthropic's latest Fable 5 and Mythos 5 systems. (CNBC)
State Department launches an “AI Opportunity Partnership” with allied nations. The U.S. State Department published a joint statement on June 25 establishing an AI Opportunity Partnership among participating governments. The signatories said they will coordinate on the physical foundations of AI—critical minerals, energy, compute, and semiconductor manufacturing—and on building trusted technology supply chains. The statement frames AI infrastructure as a shared economic and security priority. (U.S. State Department)
That's today's edition of AI Current. Reply with what you're seeing in AI, and forward this to a colleague who would find it useful. See you tomorrow.
A WORD FROM OUR SPONSOR
The #1 Time-Series Database Built on Postgres.
Columnar storage. No split architecture. No new query language. TimescaleDB extends Postgres with hypertables, Hypercore compression, and continuous aggregates so analytics run on live data.
CERN uses it to handle sensor data from the Large Hadron Collider.
One database. No pipeline.
A WORD FROM OUR SPONSOR
Be the one who already read it. TLDR is the 5-minute tech brief that 7M+ subscribers start their morning with, curated by ex-Google and Anthropic engineers.