GLM-5.2 Cut 71.9%, Nemotron-3-Super Jumps 252.9% in Busy Pricing Day

Prices: z-ai/glm-5.2 input dropped $0.3486→$0.098/MTok (-71.9%, evt 48764; also -70.5% on OpenRouter, evt 48318). DeepSeek-v4-flash saw multiple cuts: -36.5% on DigitalOcean to $0.08/MTok (evt 48920) and -11.1% to $0.08/MTok across Kilo/OpenRouter routes (evt 48906, 48904, 48902). Venice's openai-gpt-56-luna and luna-pro both fell -78.7% to $0.2667/MTok input (evt 48567, 48564). DeepSeek-v3.2 eased -3.3% (evt 46778, 46382), Kimi-k2.6 dipped -1.6% (evt 46365), and Inkling slid -5% (evt 46740, 46322).

On the up side: Nvidia nemotron-3-super-120b-a12b spiked +252.9% input to $0.3/MTok (evt 48705, 48349). Kimi-latest rose +12% to $2.8/MTok (evt 48752, 48665, 48330). Minimax-m2.7 climbed +11.1% (evt 48744, 48346). Qwen3.6-35b-a3b ticked +7.1% (evt 48712, 48610, 48331), and Kimi-K3 rose +5.6% (evt 48888).

Deprecations: inclusionai/ling-3.0-flash:free is no longer listed on OpenRouter after three absent snapshots (evt 50248).

Plans: Base44 and Zencoder posted new yearly tiers (evt 49127, 49138). Lovable dropped Pro ($25/mo) and Business ($50/mo) listed pricing (evt 49134). Devin removed its $80/mo Teams price (evt 45179). Claude Code restructured tiers, adding Team and dropping Team seats/Education (evt 49129). Amazon Q Developer renamed Free/Pro to Free Tier/Pro Tier (evt 49125). Roo Code swapped self-hosted tiers for user-count tiers (evt 49137).