DeepSeek dropped back to its old prices.

2026-09-09 02:49

DeepSeek dropped back to its old prices. Starting 12:00 noon Beijing time, September 10 — Flash series repriced: - Cache hit: ¥0.02/M tokens (back to historical low) - Input, no cache: ¥1 (July 16 pricing) - Output: ¥4 (still double the pre-hike number) Two months of whiplash: April slasher to ¥0.02 (industry floor), then August 16 peak/off-peak surge of 150%-400%, now September 10 pulling back. Price cuts are a race to the bottom — Zhipu GLM 5.3 Flash went half-price to poach users, Meta Muse Spark 1.3 pressuring from the other side.

Meta shipped Muse in the US — an AI agent that books travel

, sends emails, fills forms, does your shopping. Not a chatbot that talks. An agent that does things. Runs in its own locked-down VM with a separate browser. Hooks into email, calendar, Stripe, health apps, smart home. You control every permission. Invite only or waitlist for now. Free tier, plus $20 and $100/mo paid plans. Big claim: even Meta can't touch your data. Runs on their Muse Spark models.

2026-09-09

reasonable

2026-09-06

Meituan shipped LongCat-2.5

. 1.6T params, ~48B active, native multimodal, aimed at long-horizon agents, trained start to finish on Chinese hardware. Images get quantized into multi-level discrete tokens, and that's where Visual Coding and 3D pick up. On the RL side, failed trajectories get turned back into verifiable training tasks. W4A4 inference, 43% faster prefill. Meituan's own evals put it between Claude Opus 4.8 and Opus 5.

2026-11-02

SenseTime open-sourced SenseNova Loop-9B-Preview.

It stacks layers instead of parameters. The decoder has 32 layers, but layers 4 through 27 run twice with shared weights, so a forward pass does 56 layers of work while the checkpoint stays at 9B. Their own eval numbers: 71.38 on BIRD-Dev and 13.72 on AutomationBench, both above Qwen3.5-27B. Terminal Bench is still behind.

2026-11-02