GLM 4.6/5-series, DeepSeek Flash cut across OpenRouter; DeepSeek Pro up 153-203%

Prices Mixed moves across model catalogs today. DeepSeek v4-pro jumped hard on nano-gpt (input $0.435→$1.1/MTok, +152.9%, evt 60838, 60839) and even more via kilo ($0.435→$1.32/MTok, +203.4%, evt 60844), with deepseek-latest on nano-gpt matching the +152.9% move (evt 60376). Meanwhile DeepSeek v4-flash got cheaper across OpenRouter, kilo and nano-gpt: input $0.14→$0.0769/MTok (-45.1%, evt 60686, 60223) and a smaller cut $0.08→$0.0783/MTok (-2.2%, evt 60810, 60804, 60799).

Z-AI's lineup saw broad cuts on OpenRouter and kilo: glm-5 input $0.95→$0.6/MTok (-36.8%, evt 60698, 60569, 60248), glm-5.1 $1.4→$0.966/MTok (-31%, evt 60701, 60230), glm-5.2 $0.63→$0.49/MTok (-22.2%, evt 60712, 60208) — while glm-4.6 input rose $0.5→$0.55/MTok (+10%, evt 60715, 60576, 60255).

Gemini 3.6-flash was cut -50% across edenai and nano-gpt variants ($1.5→$0.75/MTok input, evt 60843, 60842, 60841, 60816), and gemini-flash-latest matched that cut on nano-gpt (evt 60340). Cortecs' gemini-3.5-flash rose +11.2% (evt 60829). Kimi-k2.6 input fell $0.95→$0.65/MTok (-31.6%, evt 60720, 60227). Qwen3.5-397b-a17b (-22%, evt 60671, 60246) and qwen3.5-35b-a3b (-10% input, evt 60661, 60239) also dropped. GPT-5.6-luna cut input -80% (evt 60745); gpt-5.6-terra cut -20% (evt 60746). Llama-4-maverick output rose +14.9% (evt 60728, 60271); llama-3.3-70b-instruct ticked up +1.7% (evt 60631).

Releases Deepseek-v3.1-terminus scheduled for removal from OpenRouter, expiring 2026-08-17 (evt 62210).

Plans Zencoder updated Pro, Pro Plus, Pro Max and Enterprise limits (evt 61064). Warp dropped its Free plan's listed $0/yr price and adjusted limits across Build, Max, Business and Enterprise tiers (evt 61062). Z Code updated its plan set (evt 61063).