AI Daily · Sep 14, 2026: DeepSeek V4.1 Flash launches on Command Code with 6x usage boost for GOAT plan
Command Code launches DeepSeek V4.1 Flash, adding $60 usage to GOAT plan (valid 10 days) for 6x weekly usage; MiniMax H3 video model gets community optimization—FastH3 achieves 4-step distillation and runs on Apple Silicon.
Pick a date
-
Command Code boosts GOAT plan with DeepSeek V4.1 Flash usage
Command Code now uses DeepSeek V4.1 Flash as its top model and extends its $10/month GOAT plan with $60 of usage (154K requests / 7.8B tokens) over 10 days—a 6x usage boost for a week.
-
MiniMax H3 open-weights video model gets FastH3 4-step distillation
MiniMax's open-weights H3 video generation model with native stereo audio and multimodal reference control is gaining community optimizations: FastH3 achieves 4-step distillation and now runs on DGX Spark and Apple Silicon.
-
Perplexity Portable Computer now runs locally on Windows RTX PCs
Perplexity's Portable Computer is now available on Windows PCs with NVIDIA RTX GPUs, letting users run the full local AI stack—harness, agents, and models—directly on their machine. Requires an RTX GPU with at least 24GB VRAM.
-
Warning: cheap AI inference tokens may hide data theft risks
A developer warns that suspiciously cheap AI inference tokens may involve malicious behavior such as sending fake tool calls to steal data or selling entire conversation traces, potentially exposing API keys.
-
Tencent AI open-sources CubeSandbox: lightweight agent sandbox runtime
Tencent AI open-sourced CubeSandbox, a lightweight sandbox runtime for AI agents built on RustVMM and KVM, featuring cross-node pause/resume and improved snapshot restore.
-
Perplexity Portable Computer adds local MCP and scheduled tasks
Perplexity's Portable Computer now supports local Model Context Protocol (MCP), letting users connect custom tools and app integrations, plus scheduled tasks for recurring work on their PC.
-
Sam Altman backs federal AI safety framework, warns against power concentration
OpenAI's Sam Altman states AI progress must avoid humanity losing control and excessive concentration of power, welcoming a federal framework with consistent safety requirements for frontier AI.