LLM routing news

LLM API pricing updates

Short, source-aware updates about model pricing, routing decisions, local AI alternatives, and apiroute.dev data changes. These entries are the source material for approved Bluesky and Mastodon posts.

Claude Pricing data

Claude Opus 4.8 pricing added to apiroute.dev

Claude Opus 4.8 is now included in the pricing API, model pages, comparison pages, and sitemap. The current snapshot lists $5 input, $25 output, and $0.50 cache-read per 1M tokens.

Read update
Local AI GPU fit

Before cloud routing, check whether your GPU can run the model locally

apiroute.dev now links more directly to localai.apiroute.dev, the companion calculator for VRAM fit, quantization, Ollama commands, and local-vs-cloud fallback decisions.

Read update Open GPU checker
Agent routing SLM routing

Token Waste Check: when agent workflows should route to cheaper SLMs

Repeated agent prompts do not always need a frontier model. apiroute.dev now compares selected agent routes with cheaper SLM, budget, and local-open alternatives before routine workflows spend tokens.

Read update
Partner option

AI/ML API added as a clearly labeled provider option

apiroute.dev now includes a small AI/ML API partner card in the business layer. The link is separated from the model ranking UI and does not change price tables, Best Route recommendations, or calculator output.

Editorial note: provider pages should still be verified before production routing. The AI/ML API link is an affiliate link and is marked as sponsored in the HTML.

Read disclosure
Indexing

Google processed the apiroute.dev sitemap with 187 URLs

The sitemap for apiroute.dev was processed in Google Search Console with 187 discovered URLs. Core pages, model pages, comparison pages, data disclosures, and machine-readable endpoints are now prepared for discovery.