People searching "Artificial Analysis alternatives" already know the Intelligence Index exists. They want a second view, not a definition.
Artificial Analysis runs the evals. We don't. The 11 Aug 2026 snapshot we ingest still has Claude Opus 5 (Adaptive Reasoning, Max Effort) at rank 1, score 63.1. GPT-5.6 Sol (max) is the first OpenAI row at rank 5, 60.9. If you need those numbers from the lab that produced them, stay on Artificial Analysis. If you need the same snapshot next to makers and startup credits, that is a different job. The sites below split that job in incompatible ways. Mixing them wastes a week.
What "alternative" actually means here
AA is an eval lab. Most pages that rank for this query are not eval labs. They answer a different question and borrow the word "leaderboard."
| Site | Job | Not the job |
|---|---|---|
| Artificial Analysis | Intelligence Index, speed, price | Maker directory, startup credits |
| LMSYS Arena | Blind human preference votes | The same harness as AA |
| Hugging Face Open LLM Leaderboard | Open-weight evals | Closed APIs (Opus, Sol, Gemini) |
| OpenRouter | Routing, list price, usage | Independent evals |
| Models & Makers | Dated AA snapshot, makers, perks | Running the evals |
Pick the row that matches the question you are actually asking. If the question is "who scored 63.1 on 11 Aug," the answer is still AA, and we are a copy of that file with extra columns around it.
If you want a second eval, not a prettier table
Arena answers "which model do people pick when the names are hidden?" That is not AA's question. AA's question is a fixed use. Both are useful. They disagree often enough that you should not average them in your head and call it science.
Hugging Face's open LLM board answers a third question: open weights only. On our 11 Aug snapshot, Qwen3.8 Max sits at rank 9 with 58.1 and an Open Weights label we inferred from the name. That row is the kind of thing the HF board is built for. Opus 5 is not. If your constraint is "we will not call a closed API," start there, then come back to /leaderboard only for the open-weight slice.
OpenRouter is a router. Use it when the constraint is which endpoint is up, at what list price. Do not treat its ranking UI as a substitute for AA. Usage is not an eval.
If you want the AA numbers plus the rest of the stack
We snapshot AA into content/leaderboard_data.json and publish it on /leaderboard. Weekly deltas live on /movers. How licenses get inferred, and what we refuse to claim, is on /methodology. Long-form comparisons sit on /posts. Thin /compare scorecards are noindex on purpose. They are not an "alternative" to AA. They are a leftover scorecard.
The other half of the site is not a leaderboard. /perks is a startup-credits directory. Apply links sit behind a sign-in or a confirmed newsletter email. That is the actual product difference versus AA, Arena, and HF. Not a better Intelligence Index. We do not have one of those.
Head-to-head with the lab we ingest: Models & Makers vs Artificial Analysis. How to pick a comparison tool by job: Best AI model comparison tools in 2026. Router vs eval: OpenRouter alternatives.
What we will not do
We will not invent Arena Elo or Hugging Face scores here. Those sites publish their own tables. If a number is not in the 11 Aug file or on the live board, it does not belong on this page.
We will not call ourselves "the best Artificial Analysis alternative." That sentence is ads. We ingest their board. Credit stays with them. When the snapshot file moves, this page should move with it: change the table, bump the updated date in the byline.
Outbound links above go to the real sites. If a vendor page 404s, tell hello@modelsandmakers.com.
Frequently asked questions
Is Models & Makers an Artificial Analysis alternative?
Only if your job is "dated snapshot plus makers plus credits." If your job is "run or inspect the evals," no. Use AA.
Does LMSYS Arena replace Artificial Analysis?
No. Blind votes versus a fixed use. Use both when a pick is expensive. Do not flatten them into one rank.
Where is the live board?
/leaderboard. The snapshot date is in the page chrome. If a license looks wrong, read /methodology before filing a complaint. Inference from names is a guess with a changelog, not a legal opinion.
Do you sell rankings?
No. The board is the AA ingest. Perk apply links are gated. We do not sell a rank.
Bottom line
Stay on Artificial Analysis for the evals. Use Arena for preference, Hugging Face for open weights, OpenRouter for routing. Use this site when you want that AA snapshot next to makers and credits, with the dates visible. That is the alternative. It is a narrow one.