Lambda Inference API wind-down notice (lambda.ai/inference)

Fetched 2026-10-15. The whole page body reads: 'Run inference your way — As the Inference API winds down, you can continue deploying and scaling models seamlessly on NVIDIA GPU instances.' Lambda's hosted-token product (api.lambda.ai/v1, OpenAI-compatible, served Llama / DeepSeek / Hermes / Qwen models) is being retired and users are steered to on-demand GPU instances instead. docs.lambda.ai/public-cloud/inference-api/ already 404s. Lambda's GPU-hour instance pricing is unaffected and remains in data/supply/ (supply/lambda-gpu-cloud-h100). No public token-pricing page remains reachable, so per-token rates are not recorded - the record exists to answer 'does Lambda still rent tokens' (no, winding down) rather than to quote a price.