OpenAI announcement image for GPT-6 Astra, a spiral galaxy of white, blue and amber points of light curling around a bright core on a near black star field.

GPT-6 Astra lists at $10 and $50, 2.5x what GPT-5.6 Sol costs

4 September 2026
Google's official announcement image for the release, reading Introducing Gemini 3.8 Flash and 3.8 Flash Cyber in black type over a pale blue background with a blurred white chevron and the four colour Gemini spark below.

Gemini 3.8 Flash keeps the price and the 1 January cliff

3 September 2026
Answer card stating that Anthropic announced Enterprise Frontier Safeguards on 1 September 2026, that activity data used for misuse monitoring moves into cloud storage the customer controls under the customer own encryption keys, that Anthropic charges nothing for the feature while the cloud provider bills storage and egress, and that the phased rollout starts later in autumn 2026 with interim zero data retention on Fable 5 and Fable 5.1 for eligible customers.

Anthropic moves retention into your own cloud, for 30 days

3 September 2026
Official Google diagram of a client connection in three numbered steps: a DNS lookup with a query and an address, a TLS ClientHello and ServerHello, then a content exchange with a website. A callout on the DNS step reads 25% of global web traffic is now protected by encrypted DNS, and a callout beside an Android phone on the ClientHello step reads Android 17 supports ECH GREASE by default.

Android 17 hides the SNI, not your DNS or destination

3 September 2026
Still frame from the Claude Fable 5.1 launch video showing model-designed protein binders in orange docked against twelve grey target proteins, rendered as ESMFold2 structure predictions.

Claude Fable 5.1 breaks forced tool use, cuts cache 75%

1 September 2026
Answer card stating that on 31 August 2026 the European Commission designated ChatGPT a Very Large Online Search Engine under the Digital Services Act, the first conversational AI service classified that way, because it answers user prompts and queries including by searching the web, with OpenAI having declared roughly 159.1 million average monthly users in the European Union for ChatGPT search.

The EU now calls ChatGPT a very large search engine

3 September 2026
Answer card stating that on 31 August 2026 the Department of War added OpenAI ChatGPT Mil and Starshield AI Grok for Government to the GenAI.mil portal alongside Google Gemini, all three accredited at Impact Level 5 for Controlled Unclassified Information, with 1.7 million unique users onboarded out of roughly 3 million eligible personnel, and ChatGPT Mil currently serving GPT-5.4 Terra with GPT-5.6 Terra said to be rolling out.

ChatGPT Mil and Grok reached IL5 on GenAI.mil

3 September 2026
Answer card stating that Anthropic opened a research preview of the Model Hardware Standard on 27 August 2026, standardising the driver layer between an operating system and a laboratory instrument with read and write primitives plus discovery and safety limits, reachable through MCP as well as a command line and code files, with no public specification published.

Anthropic’s Model Hardware Standard is gated, and sits under MCP

3 September 2026
Official Cohere key art for the Parse 5 launch: the Cohere mark and the wordmark Parse with a superscript 5 in white, centred on a soft out of focus gradient of deep blue, violet and amber curves.

Cohere Parse 5 is $1.50 per 1,000 pages, on three of five dimensions

3 September 2026
Title card from the OpenAI announcement video: a man sits on a blue sofa in a loft with tall windows and potted plants, a laptop open on the coffee table in front of him, with the words WebMCP in ChatGPT in large white type across the lower left.

WebMCP in ChatGPT needs GPT-5.6 Sol or Terra

3 September 2026
The Agentic Coding section of the official Hy4 preview benchmark appendix published by Tencent, a table comparing Hy3 and Hy4 preview against DeepSeek V4 Pro 0813, Qwen 3.8 Max, GLM 5.3, Kimi K3, GPT 5.6 Sol and Claude Opus 5 across SWE-bench Multilingual, SWE-bench Pro, DeepSWE, three SWE Atlas tasks, SWE-Marathon, Terminal-Bench 2.1, NL2Repo-Bench, CyberGym, ProgramBench, PostTrainBench and Harbor-Index.

Tencent’s 770B Hy4 tops one benchmark row in 46

3 September 2026
Official Qwen architecture diagram for Qwen3.8-Flash-Next, showing input tokens feeding a vocabulary embedding and a separate n-gram embedding layer at layer 2, a stack of Gated DeltaNet layers punctuated by Qwen Sparse Attention layers in a three to one hybrid block, Gated Residual read and write gates around each MoE block, and MTP modules beside the prediction head.

Qwen3.8-Flash-Next runs 6B active and weighs 173 GiB

3 September 2026
Friday, September 4, 2026
No Result
View All Result
No Result
View All Result

Tag: research

  • Trending
  • Comments
  • Latest

Welcome Back!

Login to your account below

Forgotten Password?

Retrieve your password

Please enter your username or email address to reset your password.

Log In
No Result
View All Result

Copyright © 2026 Stephane Cardon.