Choosing between frontier APIs, open weights, and self-hosting — and the trade-offs behind each.
Sign in to track your progress across sections.
20 articles
Listed inGoogle GeminiModels & Providerson
Images, audio and video travel in the request and are billed as input tokens, at different per-second rates for audio and video. The same models are reachable through the Gemini Developer API and through Google Cloud, which differ in the platform around them rather than in capability.
Listed inSelf-Hosted ModelsModels & Providerson
Sizing a GPU for a self-hosted model means budgeting for more than the weights. The KV cache grows with concurrency, and exhausting it preempts in-flight requests that are then recomputed from the start.
Listed inHugging Face Inference SDKModels & Providerson
One client in front of hosted providers, dedicated endpoints and a local server, with a two-line diff from an OpenAI client.
Listed inOpenAI GPT & o-seriesModels & Providerson
Reasoning tokens are billed as output, invisible in the API, and share your context window. That explains both the bill and the truncation.
Listed inQwenModels & Providerson
Apache 2.0, over 100 languages, dense and mixture-of-experts across a wide size range. Why tokenizer coverage shows up in your bill.
Listed inModels on Hugging FaceModels & Providerson
A model card is a data structure. Read the license first, then what it was derived from, then whether that benchmark number has a source.
Listed inChoosing the Right ModelModels & Providerson
Compare models on capability, latency, cost per request, context window, maximum output, and where they can run. Price per million tokens is not cost per request, and twenty saved examples from your own application settle the choice faster than a leaderboard.
Listed inCohereModels & Providerson
Reranking scores the query and document together instead of comparing two vectors, which is why it goes second and not first.
Listed inOpenRouterModels & Providerson
One endpoint makes changing your mind cheap. The training opt-out governs the providers, not the router, and retention is not routed on at all.
Listed inMeta LlamaModels & Providerson
Downloadable weights come under a custom licence rather than Apache or MIT. It requires a "Built with Llama" notice, a model name beginning with Llama, and a separate agreement above 700 million monthly active users.
Listed inOllamaModels & Providerson
One command pulls a packaged build and serves it, which means a quantization was chosen for you. Compatible client surfaces let existing code point at it, and a single-user local runner is no evidence about serving traffic.
Listed inTransformers.jsModels & Providerson
Running transformer models in the browser with WebGPU and WASM.
Listed inAnthropic ClaudeModels & Providerson
The Claude family, its long-context and tool-use strengths, and the Messages API shape.
Listed inGemmaModels & Providerson
Google's small open models, sized for laptops and edge deployment.
Listed inHugging Face TasksModels & Providerson
Task taxonomy as a lookup table from problem to candidate models.
Listed inClosed vs Open Source ModelsModels & Providerson
What you give up and gain by leaving a hosted frontier API.
Listed inDeepSeekModels & Providerson
Open reasoning models that closed much of the gap with frontier APIs.
Listed inHugging Face HubModels & Providerson
The registry for models, datasets, and spaces, and how to navigate it.
Listed inMistralModels & Providerson
Small, fast, permissively licensed models and mixture-of-experts variants.
Listed inLM StudioModels & Providerson
A desktop GUI for pulling, comparing, and serving local models.