{"type":"post","id":"hadley-wickham-estimates-what-it-costs-to-serve-073584","slug":"hadley-wickham-estimates-what-it-costs-to-serve-073584","title":"Hadley Wickham estimates what it costs to serve Kimi K3 through hosted APIs versus self-hosting","description":"Hadley Wickham estimates what it costs to serve Kimi K3 through hosted APIs versus self-hosting, using provider prices and a rough GPU-server model.","retrieval_nugget":"Hadley Wickham estimates what it costs to serve Kimi K3 through hosted APIs versus self-hosting, using provider prices and a rough GPU-server model. His calculation suggests API pricing can leave room for operating margin, while the clearest subsidies may sit in subscriptions and model training. For teams budgeting AI, the assumptions matter as much as the headline cost.","published_at":"2026-11-05","updated_at":"2026-11-05","record_date":"2026-11-05","date_kind":"published_at","topics":["inference","pricing"],"entities":["Hadley Wickham","Kimi K3","OpenRouter","Moonshot AI","SemiAnalysis","Posit Assistant","Opus 5.5"],"related_patterns":[],"novelty_label":"new_fact","what_changed":"Adds a provider-by-provider Kimi K3 cost comparison, including blended agentic-coding rates, to the prior coverage of steep LLM API price cuts.","prior_refs":["https://newruntime.com/posts/a-dive-into-why-openai-anthropic-cut-llm-8966c7/"],"source_url":"https://substack.hadley.nz/p/how-much-do-tokens-really-cost","source_urls":["https://substack.hadley.nz/p/how-much-do-tokens-really-cost"],"schema_version":"newruntime-agent-readable-v0.2","stable_id":"post:hadley-wickham-estimates-what-it-costs-to-serve-073584","status":"published","visuals":[],"editor":{"role":"editor","name":"Andrey Reshetnikov","profile":"https://newruntime.com/owner-profile.md","profile_json":"https://newruntime.com/owner-profile.json","same_as":["https://www.linkedin.com/in/reshetnikov1/","https://t.me/lapetuse"]},"routes":{"html":"https://newruntime.com/posts/hadley-wickham-estimates-what-it-costs-to-serve-073584/","markdown":"https://newruntime.com/posts/hadley-wickham-estimates-what-it-costs-to-serve-073584.md","json":"https://newruntime.com/posts/hadley-wickham-estimates-what-it-costs-to-serve-073584.json"},"markdown_url":"https://newruntime.com/posts/hadley-wickham-estimates-what-it-costs-to-serve-073584.md","next_reads":[{"type":"topic","path":"/topics/inference/","url":"https://newruntime.com/topics/inference/","title":"Inference","reason":"Explore the inference topic hub.","media_type":"text/html","markdown_url":"https://newruntime.com/topics/inference.md","json_url":"https://newruntime.com/topics/inference.json"},{"type":"topic","path":"/topics/pricing/","url":"https://newruntime.com/topics/pricing/","title":"Pricing","reason":"Explore the pricing topic hub.","media_type":"text/html","markdown_url":"https://newruntime.com/topics/pricing.md","json_url":"https://newruntime.com/topics/pricing.json"},{"type":"collection","path":"/posts/","url":"https://newruntime.com/posts/","title":"Field Notes","reason":"Open the latest editorial analysis.","media_type":"text/html","json_url":"https://newruntime.com/posts.json"},{"type":"collection","path":"/sources/","url":"https://newruntime.com/sources/","title":"Sources","reason":"See the original sources this archive cites.","media_type":"text/html"}]}