Gemini 3.7 Flash doubles to $1.5/$7.5 per MTok; Replit Core cut to $20/mo

Prices Google/Vertex Gemini 3.7 Flash aliases doubled uniformly to $1.5 input/$7.5 output per MTok across edenai (evt 68598,68597,68596,68595,68594,68593) and kilo (evt 68587).

DeepSeek v4 gateways diverged: edenai's v4-flash rose to $0.44/$1.32 (evt 68592) and v4-pro to $1.32/$3.96 (evt 68581); merge-gateway's v4-pro hit $0.66/$1.98 (evt 68578,68308) and v4-flash $0.22/$0.66 (evt 68567,68307); llmgateway instead cut v4-flash to $0.05/$0.09 (evt 68238).

GLM-5.2 split by gateway: OpenRouter batch doubled to $1.4/$4.4 (evt 68558), OpenRouter spot rose to $0.966/$3.036 (evt 68458); vercel cut to $0.8/$2.55 (evt 68371), hyper cut to $1.1/$3.851 (evt 68334), ambient cut to $0.6/$2 (evt 68245).

Kimi-k2.5 fell to $0.45/$2.25 on models.dev (evt 68463) but rose to $0.5744/$3.005 on hyper (evt 68323); Kimi-k2.7-code batch doubled to $0.95/$4 (evt 68559).

nano-gpt cut GPT-latest, gpt-5.6-sol/sol-pro, and gpt-chat-latest 50% to $2.5/$15 (evt 68234,68233,68232,68231).

OpenRouter batch models minimax-m3 ($0.3/$1.2, evt 68561), nemotron-3-ultra ($0.6/$3.6, evt 68560) and thinkingmachines inkling ($1/$4.05, evt 68555) all doubled; llmgateway trimmed nemotron-3-ultra output to $2.2 (evt 68243).

Qwen models mostly cut: qwen3-next-80b -10% (evt 68417), qwen3.5-122b -10.3% (evt 68415), qwen3.6-27b -50% (evt 68400), gemma-4-31b -10% (evt 68380); but qwen3.8-27B-TEE on chutes rose +12.5% (evt 68617), qwen3-coder-480b on llmgateway rose +26.7% (evt 68237), minimax-m2.7 on hyper rose +14.6% (evt 68333), and glm-5.1 on hyper fell -10.8% (evt 68314).

Plans Replit cut Replit Core to $20/mo from $25, trimming monthly credits to $20, and now lists a $0/yr Starter tier (evt 68832).