Loading current prices…
About 750 words ≈ 1,000 tokens. Re-used project context is billed at the cheaper cached rate where the model offers one — that is why AiArena carries accepted context forward instead of re-sending everything.
Only from your models (⭐ in the list). The cheapest one that is good enough for the kind of task wins — proven by your own AI-duel and accepted results once you have some, otherwise by its price tier. House AI turns included in your plan count as already paid for.
| ⭐ | Model | Read $/M | Write $/M | Cached $/M | Context | This job | Tier |
|---|
Cheap scout first (sort, classify, extract, format) → a mid-tier maker when real work is needed → a frontier expert only where your own results show it is worth the extra. Same result, lower cost. List prices only: reasoning tokens, tools, search, images and very long prompts can add to the real bill.