1 credit ≈ $0.000001 · llmx bills 20% of the provider's API price · suppliers earn 50% of what consumers pay ·
| Provider | Model | API input / M | API output / M | llmx input / M | llmx output / M | llmx vs API | Status |
|---|---|---|---|---|---|---|---|
| Claude | claude-opus-5 | $5.00 | $25.00 | $1.00 | $5.00 | -80% | live |
| Claude | claude-fable-5-1 | $10.00 | $50.00 | $2.00 | $10.00 | -80% | live |
| Claude | claude-opus-5-5 | $4.00 | $20.00 | $0.80 | $4.00 | -80% | live |
| Claude | claude-sonnet-5-5 | $2.00 | $10.00 | $0.40 | $2.00 | -80% | live |
| Claude | claude-haiku-5-5 | $0.10 | $0.50 | $0.02 | $0.10 | -80% | live |
| Claude | claude-sonnet-5 | $2.00 | $10.00 | $0.40 | $2.00 | -80% | live |
| Claude | claude-fable-5 | $10.00 | $50.00 | $2.00 | $10.00 | -80% | live |
| Claude | claude-opus-4-8 | $5.00 | $25.00 | $1.00 | $5.00 | -80% | live |
| Claude | claude-haiku-4-5 | $1.00 | $5.00 | $0.20 | $1.00 | -80% | live |
Provider API: pay per token, needs your own provider account. Subscription: flat monthly fee with usage caps per window. llmx: pay per token at 20% of the provider's API rate, no monthly fee, no provider account; the supplier whose subscription served you earns 50% of that. Prompt caching is priced as the provider prices it: of the input, tokens written to the cache cost 125% of the input rate and tokens read back from it 10%, so a long coding session, which resends its context every turn, is billed mostly at the read rate.
Subscription $ / M is an estimate: providers publish caps as 5-hour and weekly windows, not tokens. Edit the tokens-per-month estimate to match your own usage. llmx $ / M is blended at 9 input tokens per output token, typical for coding agents.
| Claude subscription | $ / month | Est. tokens / month | ≈ $ / M | llmx $ / M (claude-opus-5) | Subscription vs llmx |
|---|---|---|---|---|---|
| Claude Pro | $20.00 | M | $1.00 | $1.40 | -29% |
| Claude Max | $100.00 | M | $1.00 | $1.40 | -29% |
| Claude Ultra | $200.00 | M | $1.00 | $1.40 | -29% |