Netlify reprices dozens of OpenRouter models in AI Gateway, catalog grows to ~167

Netlify’s AI features pricing docs (docs.netlify.com) showed widespread repricing across the OpenRouter-routed portion of its AI Gateway catalog between the 2026-08-25 and 2026-08-26 captures. Several already-listed models moved by more than 10%: deepseek/deepseek-v4-flash-0731 input fell from $0.07 to $0.03 and output from $0.18 to $0.08 per 1M tokens (roughly halved); deepseek/deepseek-v4-pro-0813 output fell from $3.96 to $2.60 while cache-read rose from $0.04 to $0.10; moonshotai/kimi-k2.6 output rose from $2.47 to $3.39; qwen/qwen3-235b-a22b-thinking-2507 input rose from $0.23 to $0.30 and output from $2.30 to $3.00; and xiaomi/mimo-v2.5-pro output fell from $1.50 to $1.17.

The catalog also churned: 11 models were added (nine new Mistral models — codestral-2508, devstral-2512, ministral-3b/8b/14b-2512, mistral-large-2512, mistral-medium-3/3.1/3-5 — plus tencent/hy-mt2-7b and upstage/solar-pro4) and 5 were removed (deepcogito/cogito-v2.1-671b, google/gemma-3n-e4b-it, inclusionai/ling-2.6-1t, inclusionai/ling-2.6-flash, inclusionai/ring-2.6-1t), taking the roughly 161-model OpenRouter lineup to roughly 167. Every OpenRouter model still bills through Netlify’s identical 180-credits-per-$1 AI inference meter, so these are real, bill-relevant moves for any team routing traffic to the repriced models — even though Netlify’s own plan prices (Free $0, Personal $9, Pro $20, Enterprise Custom) and the 180-credits-per-$1 conversion were unchanged. Sampled Anthropic, Gemini, and OpenAI rates also held steady. Source: docs.netlify.com AI features pricing page (screenshot diff, 2026-08-25 vs 2026-08-26 captures).