How Profitable Is LLM Inference? Doing the Math on Kimi K3
The article examines the economics of running large language model inference, using Kimi K3 as a case study. It explains how token pricing, GPU rental costs, batch size, and the number of GPUs interact to determine profitability. The analysis highlights the trade‑off between latency and cost on the Pareto frontier for optimal pricing decisions.
- ▪Kimi K3, a 2.8‑trillion‑parameter model, requires at least eight B300 GPUs or sixteen B200 GPUs to fit its 1.4 TB of quantized weights.
- ▪OpenRouter charges $3 per million input tokens, $15 per million output tokens, and $0.30 per million cached tokens for Kimi K3.
- ▪Token throughput per GPU hour is calculated as (batch size ÷ token latency) ÷ number of GPUs × 3600, linking batch size and GPU count to both speed and cost.
- ▪The Pareto frontier describes configurations where improving latency inevitably raises token cost, guiding providers in pricing fast versus regular inference tiers.
Hacker News (AI / LLM) files mainly under ai. We currently carry 2,724 of its stories.
Story provenance
Source · retrieval · rights · ranking — open for full record
inspect →
Attribution is not the same as permission. This drawer separates discovery metadata, excerpts, WeSearch-generated summaries, reuse status, and whether the publisher receives the visit. Nothing here claims a legal grant the publisher has not made.
Record
| Original publisher | Moncef Abboud |
| Canonical URL | https://cefboud.com/posts/ai-inference-costs-profit/ |
| Publication time | Wed, 29 Jul 2026 08:59:58 +0000 |
| Retrieval time | 2026-07-29T09:01:01.371Z |
| Last seen | 2026-07-29T09:01:01.371Z |
| Headline source | Publisher (no WeSearch rewrite) |
| Excerpt source | publisher body |
| Excerpt method | First ~120 words (~800 chars) of extracted publisher body, fair-use limited. |
| Summary | WeSearch · cerebras-chat (WeSearch summarizer) |
| Summary source text | contentText |
| Citation coverage | Summary is a WeSearch-generated derivative; primary citation is the original publisher URL. |
| Cluster | FhrNJT-ST4tm · 1 stories |
| Cluster logic | Grouped by semantic title/content similarity across sources within a rolling window. Same-publisher template collisions are excluded from coverage comparison. |
| Ranking reason | Story pages are not engagement-ranked. Hub feeds use recency, with optional source-diversified chronological ordering (cap consecutive stories per source). No personalized ranking. |
| Publisher visit | Yes — open original |
| Substitutes article? | No — link-out required for full text |
Rights status (four layers)
WeSearch handling by dimension
| Indexing | May the item be indexed (stored, ranked, made findable)? | Allowed |
| Snippet | May a short excerpt of the publisher's text be shown? | Allowed |
| AI summary | May WeSearch generate its own short summary of the article? | Limited |
| Retrieval / RAG | May the content be exposed for third-party retrieval-augmented generation? | Not asserted |
| Model training | May the content be used to train AI models? | Not asserted |
| Commercial reuse | May the content be reused commercially? | Not permitted |
Basis: Derived from the published RSS/Atom feed. Contact: [email protected]. Reviewed: 2026-07-24.
Opening excerpt (first ~120 words) tap to expand
How Profitable is LLM Inference? Doing the Math on Kimi K3A look at LLM inference economics (batch size, GPU count, and the Pareto frontier that sets token prices) applied to Kimi K3 with back-of-the-envelope math. Posted Jul 29, 2026 By Moncef Abboud | 5 min read How Profitable is LLM Inference? Doing the Math on Kimi K3 Contents How Profitable is LLM Inference? Doing the Math on Kimi K3 IntroHow profitable is LLM inference?This is a central question sitting at the heart of the AI mania.In this post, we’ll cover token pricing basics, build up some intuition, then apply it to the recently released Kimi K3 to answer the profitability question.LLM Pricing TLDRTo produce tokens, we need GPUs and model weights. We load the weights onto one or many GPUs.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at Moncef Abboud.
Discussion
0 commentsMore from Moncef Abboud
