Which model to choose

The extension translates via OpenRouter — pay per token. Here are tested models, sorted from cheapest to most expensive.

Model ~Price per episode Quality Best for
Llama 4 Maverick Free Basic Try without paying. Rate limit: 20 req/min, 200/day.
DeepSeek V3.2 ~$0.007 Good Cheapest paid option. Excellent value for money.
Gemini 3 Flash ~$0.04 Good+ Slightly better quality than Lite. For those who want more accuracy.
Claude Haiku 4.5 ~$0.06 Excellent Anthropic quality at an accessible price. Catches humor, slang, context.
Claude Sonnet 4.6 ~$0.19 Best Maximum translation quality. For those who care about every nuance.

Prices are approximate, per ~40 min episode (~500 subtitle lines). Calculation: ~12K input + ~10K output tokens.

You can enter any OpenRouter model in the extension — not just the ones listed here. Just switch to "Custom model" and enter the model ID.


What to choose?

Try for free
Llama 4 Maverick
Zero cost, decent quality. Has rate limits.
Watch regularly
Gemini 3.1 Flash Lite
2 cents per episode — a full season costs 20 cents. Fast and good quality.
Quality matters
Claude Haiku 4.5 or Sonnet 4.6
Anthropic models best understand context, humor, and cultural references.

How pricing works

  • OpenRouter registration is free. You top up your balance and pay only for tokens used.
  • A token is roughly ¾ of a word. An episode with 500 subtitle lines uses ~22K tokens (input + output).
  • Repeat viewings are free — translations are served from cache.

Install the extension and choose a model in settings