Author

Shan

Shan edits AI Mastery's daily reporting and evaluates new models, developer tools, and infrastructure through an engineering-first lens.

S
ShanEditor & AI Systems Builder

Shan edits AI Mastery's daily reporting and evaluates new models, developer tools, and infrastructure through an engineering-first lens.

AI PlatformsAutomationLLM OpsDeveloper Tools

Recent work

news
Shan2026-09-10
DeepSeekMixture of ExpertsLong ContextOpen WeightsInference Efficiency

DeepSeek-V4.1-Flash: 890 Bytes Per Token, 437x Smaller KV Cache

DeepSeek's 552B MoE model cuts global KV cache to 890 bytes per token — 437x below V1 — using CED, CSA2, and FP4 quantization.

Read more
news
Shan2026-09-10
OpenAIGPT-6Agentic AIAPI PricingCybersecurityBenchmarks

GPT-6 Astra: $10/M Tokens, 57.9% Terminal-Bench, Critical Cyber Flag

OpenAI's GPT-6 Astra scores 57.9% on Terminal-Bench 4.0, costs $10/$50 per million tokens, and is the first model to hit the Critical cybersecurity threshold.

Read more
news
Shan2026-09-10
Time SeriesOpen WeightsIBMForecastingGIFT-Eval

IBM's 385M-Param PatchTST-FM-r2 Ranks 2nd on GIFT-Eval Zero-Shot

IBM's Granite PatchTST-FM-r2 ranks 2nd zero-shot on GIFT-Eval under Apache 2.0/OpenMDW 1.0, beating larger pretrained rivals.

Read more
news
Shan2026-09-10
OpenAIAI SafetyAI GovernanceAlignmentFrontier AI

Paul Christiano, RLHF Co-Inventor and AI Doomer, Joins OpenAI Board

The Alignment Research Center founder joins OpenAI's Safety and Security Committee, which holds final authority over model releases like Astra.

Read more
news
Shan2026-09-09
Open SourceSecurityAI AgentsGoogleVulnerability Research

Google's Mantis Cuts AI Security False Positives With Sandbox-First Pipeline

Google open-sources Mantis, a slash-command toolkit that grounds AI vulnerability findings in sandbox execution, cutting token overhead by over 85%.

Read more
news
Shan2026-09-09
Amazon BedrockOpenAIAgentic AIEnterprise AICybersecurityOpen Weights

GPT-6 Astra Hits Amazon Bedrock With 1M-Token Context and Critical Security Tier

GPT-6 Astra is now generally available on Amazon Bedrock with a 1M-token context window, chip-level operator isolation, and OpenAI's first Critical cybersecurity classification.

Read more
news
Shan2026-09-09
MetaAI AgentsAgentic AISecurityMuse SparkOpen Weights

Meta Muse Runs Each User's Agent in a Dedicated Secure Cloud VM

Meta's Muse gives every user an isolated cloud VM, a Sentinel control plane, and surrogate credentials — with a $130,000 bug bounty on injection attacks.

Read more
news
Shan2026-09-09
Mistral AISovereign AIEuropean AIFundraisingSamsung Electronics

Mistral's €3B Series D Targets 1 GW of European Compute by 2030

Mistral AI closes a €3B Series D at a €21B+ valuation, led by Samsung, targeting 1 GW of European compute capacity by 2030.

Read more
news
Shan2026-09-09
ArchitectureInference EfficiencyAmazon SageMakerOpen WeightsBenchmarks

150M-Parameter BDH-CQ Scores 29.2% on ARC-AGI-1 at $0.0007 per Task

Pathway's 150M-parameter BDH-CQ model scores 29.2% pass@2 on ARC-AGI-1 at $0.0007 per task, trained on SageMaker HyperPod with H200 GPUs.

Read more
news
Shan2026-09-09
Open WeightsAmazon SageMakervLLMMixture of ExpertsInferenceQuantization

2.4T-Parameter Qwen3.8 Runs on One Node With vLLM and NVFP4

AWS documents a single-node serving path for Qwen3.8-2.4T-A95B on a p6-b300 instance using NVFP4 quantization, cutting TTFT by nearly 60%.

Read more
news
Shan2026-09-09
Amazon SageMakerMLOpsAWSFeature StoreData Engineering

SageMaker Feature Store's UpdateRecord Ends Read-Modify-Write Cycles

AWS ships UpdateRecord for SageMaker Feature Store: atomic partial writes on up to 100 features, no GetRecord required, available in all regions today.

Read more
news
Shan2026-09-09
AI InfrastructureUnikernelsFirecrackerSandboxesKubernetesOpen Weights

Unikraft Fits 1,024,132 Firecracker VMs on One 48-Core Server

Unikraft's unikernel microVMs cold-resume in ~10 ms from snapshots, enabling over 1 million scale-to-zero sandboxes on a single server.

Read more