AI Economics
MiniMax Prices M3 Cache Reads at $0.06 per 1M Tokens, Half Its Standard Rate
This June 2026 MiniMax M3 pricing note examines the cache-read line for workloads that repeatedly reuse input context. Its historical example compares a $0.06 per-million cached-read rate with a $0.12 standard rate; it is not a current September quote. The overall saving depends on how much input actually receives the cache rate and which other charges stay unchanged. Keep the workload fixed and verify the applicable provider rate before carrying the example into a new budget.
MiniMax prices cached input reads on its M3 model at $0.06 per 1M tokens, which its pricing page lists as a permanent half-off discount on a standard cache-read rate of $0.12. For a workload that reuses a long context across many calls, that line is worth checking against the input rate.
Quick answer
MiniMax reads cached input tokens for M3 at $0.06 per 1M tokens, half its standard cache-read rate of $0.12 per 1M tokens. The model’s input price is $0.30 per 1M tokens and its output price is $1.20 per 1M tokens. For a loop that reuses context, the cache-read rate is often the line that dominates, so the discounted rate is worth modeling against the input price rather than read off the headline number.
What to watch
Cached reads bill at $0.06 per 1M tokens, against the $0.12 standard rate, the $0.30 input rate, and the $1.20 output rate. For a loop that re-sends the same instructions, schema, and history on every step, cached reads can make up a large share of the token bill, so the cache-read rate can move the total more than the headline input line. Drop the rates into the ByteCosts AI cost calculator with your own context-reuse ratio to see which one your workload actually hits.
Key takeaways
- MiniMax-M3 reads cached input at $0.06 per 1M tokens, half its standard $0.12 rate, a discount its pricing page lists as permanent.
- The model’s input price is $0.30 and its output price is $1.20 per 1M tokens.
- For cache-heavy loops, compare models on the cache-read rate, not just the headline input line.
Sources
- MiniMax pay-as-you-go pricing - MiniMax, accessed June 10, 2026.
- ByteCosts AI Provider Pricing Index - ByteCosts, accessed June 10, 2026.
MiniMax Prices M3 Cache Reads at $0.06 per 1M Tokens, Half Its Standard Rate. ByteCosts. Updated 2026-09-05. https://bytecosts.com/blog/minimax-m3-cache-read-cut/