Skip to content

Archive

October 20, 2026

36 events

Importance: monitoring Anthropic Business and investment 1 source

Anthropic struggles to secure cloud capacity, according to leaked document

Anthropic is struggling to obtain enough capacity from AWS, Google and Microsoft, according to a leaked S-1 document. It is therefore leasing its own data centers and bringing in other operators, which it warns may be less reliable.

Importance: monitoring Amazon Hardware 1 source

i14 and AWS describe changes to Laplace 3B for AWS Trainium2

The companies i14 and AWS described porting Laplace 3B to AWS Trainium2 using the beta version of TorchNeuron Native. According to the authors, training reached approximately 120 000 tokens per second; the recurrent state occupies 29.5 MB per sequence regardless of context length.

Importance: monitoring Google Tools and apps update ✓ 2

Gemini assistant in Android Auto is set to check flight statuses and navigate using photos

According to Google, Personal Intelligence lets the Gemini assistant check a flight's status using a confirmation in Gmail or navigate to a location shown in a photo. The rollout to Android Auto is beginning, with cars featuring Google built-in expected to follow.

New: Supported sources of personal data also include Maps, Photos, YouTube and Search.; Gemini can retrieve details from a flight confirmation in Gmail to check the flight status.; Gemini enables navigation to a location shown in a user's photo.

Importance: monitoring New models 1 source

Elevenlabs releases two Eleven Realtime models for voice conversations

Elevenlabs has made Eleven Realtime v1 Mini and Eleven Realtime v1 Max available through ElevenAgents and the Realtime API. According to the company, the Max variant has a response time of 0.57 seconds. Billing combines a per-minute rate with token costs.

Importance: important DeepSeek Open-source update ✓ 2

DeepInfer framework supports both Nvidia SM100 and Huawei Ascend 950 chips

The DeepInfer framework for DeepSeek V4.1 Flash supports offline inference on Nvidia SM100 GPUs and Huawei Ascend 950 NPUs. DeepSeek released its previously internal software under Apache License 2.0.

Part of an overview of multiple AI topics; only this event has been covered.

New: DeepInfer supports offline inference on Nvidia SM100 GPUs.; DeepInfer supports offline inference on Huawei Ascend 950 NPUs.

Importance: monitoring OpenAI Tools and apps 1 source

OpenAI rolls out conversation locking in ChatGPT

OpenAI is gradually rolling out the Locked Chats feature in ChatGPT. Conversations can be protected with a four-digit PIN or with Face ID or Touch ID. However, locking does not change data retention or the settings for using data for training.

Importance: monitoring AI agents 1 source

Smartsheet releases MCP server to connect project data with Amazon Quick and Kiro

Smartsheet MCP Server is generally available and offers over 40 tools for reading and editing project data. Amazon Quick and Kiro connect as MCP clients; the integration described also enables automatic task status updates after a code change is recorded.

Importance: monitoring OpenAI Coding 1 source

AWS described fine-tuning GPT-OSS-20B for multilingual reasoning

AWS published a procedure for fine-tuning GPT-OSS-20B using SFT and GRPO on Amazon SageMaker HyperPod. In the test conducted by the authors, the base model followed the requested reasoning language in only 22 % of 50 examples.

Importance: monitoring Amazon Tools and apps 1 source

AWS makes GPT models available through Amazon Bedrock from Canadian regions

GPT-5.6, GPT-6 and GPT-6.1 Sol models are available through Amazon Bedrock from the Canada (Central) and Canada West (Calgary) regions. Requests are routed through cross-Region inference; the selected profile determines where processing can take place.

Importance: monitoring Security 1 source

AWS and Keyfactor described AI agent authentication using X.509 PKI

AWS and Keyfactor presented a reference architecture for agent authentication and permission delegation. It combines signed requests, user delegation, tokens restricted to the target service and certificate revocation checks.

Importance: monitoring Image generation 1 source

Adobe added an AI feature to open closed eyes in the Lightroom mobile app

The new Open Eyes feature in the Lightroom mobile app corrects closed eyes using generative AI and reference photos of the same person. It is available for iOS, iPadOS and Android; the free version offers two tries.

Importance: monitoring AI agents 1 source

Saint-Gobain used Microsoft Discovery to research cement admixtures

Saint-Gobain says it used the Microsoft Discovery platform to create a knowledge base from 62 000 patents and 44 000 abstracts in a few hours. It estimated that manual preparation would take a year; the team developed hypotheses about new admixture components in approximately six weeks.

Importance: monitoring New models 1 source

Cloudflare has made Clairic available for extracting data from documents

Clairic is available in preview on the Workers AI platform. It processes the requested fields from a document in a single pass and returns values with confidence levels; for missing data, it returns null.

Importance: important Mistral New models update ✓ official

Transcription with Voxtral 3 models starts at 0.0035 USD per minute, SDK uses LiveKit

Transcription pricing in the Voxtral 3 model family starts at 0.0035 USD per minute. The SDK for voice agents uses LiveKit, and according to Mistral AI, speech synthesis adapts emotion, pace and emphasis to the context of the conversation.

New: Transcription pricing starts at 0.0035 USD per minute.; The SDK is based on LiveKit.; Speech synthesis supports emotional adaptation.