Llm Books vs Sakana AI

The feature sets do not overlap in any dimension a buyer evaluates. Llm Books delivers pedagogy: LangChain Chains/Agents/Callback module breakdowns, LlamaIndex index internals with an enterprise knowledge-base build, a three-stage RAG treatment (indexing, retrieval, generation), Embedding fundamentals with a document-QA bot, HuggingFace/transformers plus a HuggingGPT multimodal exercise, Prompt engineering, LLMOps split into model, prompt and narrow LLMOps layers, and evaluation/testing for models, Agents and RAG. It also covers domestic Chinese model APIs — MiniMax, 智谱 AI, MoonShot — with a six-vendor capability comparison. Sakana AI delivers running infrastructure: Sakana Fugu orchestrates frontier models across multi-step tasks, Sakana Namazu (standalone API since August 2026) is a Japanese-specialised LLM, Sakana Marlin generates analyst research reports, Sakana Chat surfaces both Fugu and new-generation Namazu, and Sakana Translate does translation and proofreading on the new-generation Namazu model. Recent releases — Fugu-Ultra v1.1 with Claude Code interface support, Fugu-Cyber for cybersecurity, and a Gemma 4-trained Fugu conductor validated in testing — show the stack diversifying away from a single upstream provider, plus NVIDIA and Claude Code integrations. One is something you read; the other is something you deploy. Data-residency and export-control compliance for Japan, and Sakana's finance work with SMBC Group, MUFG and Daiwa Securities, have no counterpart in a Markdown book.

Llm Books is free, full stop — it is an open-source ebook the author explicitly frames as personal notes with a request to lower expectations and file issues for errors. There is no tier, no seat, no upsell; the only cost is your time and the risk that framework interfaces have moved past what the notes describe. Sakana AI publishes nothing. Pricing is contact-only, and the product page states plainly that it is not for companies needing self-serve signup and published prices before talking to sales, nor for buyers whose procurement cannot accommodate a custom enterprise agreement. That usually means annual contracts, minimum commitments and legal review rather than a credit card checkout. The practical consequence: these two never meet on a budget line. A solo developer can consume Llm Books today at zero cost and treat it as a syllabus; a procurement team evaluating Sakana AI is negotiating data-residency terms, export-control guarantees and support SLAs, where the licence fee is one line in a much larger agreement. Comparing a $0 download to a negotiated enterprise contract tells you nothing useful about value — they are simply different categories of spend.