People searching OpenRouter alternatives already know the product exists. They want a second path, not a definition.
OpenRouter routes requests to model APIs. It shows list price and usage. That is a real job. It is not an eval lab. Mixing those jobs is how teams pick a model from a traffic chart and then wonder why the coding agent still fails.
This page splits the field by job. We do not invent OpenRouter prices, markups, or volume ranks. Those move on their site. If you need the current rate card, read OpenRouter. If you need independent scores, you need a different URL.
Pick by job, not by "leaderboard"
| Job | Use | Skip if |
|---|---|---|
| Routing, list price, which endpoint is up | OpenRouter | You think usage is an eval |
| Independent eval (intelligence, speed, price views) | Artificial Analysis | You only need an API gateway |
| Blind human preference | LMSYS Arena | You need a fixed harness |
| Open-weight only evals | Hugging Face Open LLM Leaderboard | You are buying a closed API |
| Dated AA snapshot + makers + startup credits | /leaderboard and /perks | You need to inspect the harness |
There is no row for "best AI ranking website." That query is five jobs glued together. Glue them and you get a useless average.
If the constraint is routing
Stay on OpenRouter, or another router you already trust. We are not a router. We do not proxy tokens. We do not claim cheaper inference. Price-comparison blogs that scrape list rates are closer cousins than we are.
When someone says "OpenRouter alternative" and means "I need an endpoint today," do not send them to Arena. Votes will not keep the API up.
If the constraint is quality
Start at Artificial Analysis. They run the evals. We reprint a dated snapshot on /leaderboard. On the 16 Aug 2026 ingest, Claude Opus 5 is rank 1 at score 63. GPT 5 6 Sol is rank 5 at 61. Those are integers from content/leaderboard_data.json (updated_at 2026-08-16). Older journal posts still quote the 11 Aug file (Opus 5 Max Effort 63.1, Sol max 60.9). If the live table moved, trust the live file.
We did not produce those scores. Method, including non-claims: /methodology.
OpenRouter's ranking UI is usage. Usage is interesting. It is not AA's Intelligence Index. Do not average them.
If the constraint is votes or open weights
Arena answers "which model do people pick when names are hidden?" Hugging Face's open board answers open weights. Closed APIs (Opus, Sol, Gemini) are out of scope there. Our license labels are inferred from names when the API omits a field. Treat them as a scan aid.
If the constraint is credits and makers
That is this site. /perks is a startup-credits directory. Sign in or confirm the newsletter email to see apply links. /makers is a listing, not an eval. Weekly deltas: /movers.
Sibling pages: Artificial Analysis alternatives, Models & Makers vs Artificial Analysis, Best AI model comparison tools in 2026.
What this comparison is not
It is not "we beat OpenRouter on price." We have no OpenRouter invoice.
It is not a second router. If you need failover across vendors, that is still a routing product.
It is not a fourth page for openrouter vs artificial analysis. That matrix row is unshipped until we have a weekly GSC snapshot with BOFU impressions. One intent per URL.
Frequently asked questions
Is OpenRouter an eval lab?
No. It routes. Evals are Artificial Analysis, Arena, Hugging Face, and similar labs. Mixing the two wastes a week.
Do you publish OpenRouter prices?
No. Prices live on openrouter.ai. Inventing a rate here would be a lie.
When should I use Models & Makers instead?
When you already have a router (or a direct API) and you also need a dated AA snapshot, movers, makers, or credits. Not when you need the pipe itself.
Where do the model scores on this site come from?
Artificial Analysis. We ingest. We do not run the use. See /methodology.
Bottom line
OpenRouter alternatives depend on the job. Router → OpenRouter or another gateway. Evals → AA. Votes → Arena. Open weights → Hugging Face. Dated snapshot plus credits → us. Keep those jobs apart.
Sign in or confirm the newsletter email at https://modelsandmakers.com/perks to see apply links and browse startup credits.