Together raises Kimi K3 cached-input price to $0.30 per 1M tokens
In the serverless inference pricing table, Together raised the Kimi K3 cached input rate from $0.27 to $0.30 per 1M tokens.
Together’s pricing page · Change Detected
WHY IT MATTERS
A 3-cent move on a cached-token rate is easy to dismiss, but it's one more data point that Together treats Kimi K3 pricing as a live dial, not a fixed number — three changes to the same line in two weeks is unusual cadence even for a company that ships pricing updates about weekly. Watch whether the standard (non-cached) Kimi K3 rates move next, or whether this settles once the Oct 12 promo fully runs off.
BY THE NUMBERS
- Kimi K3 cached input: $0.27 → $0.30 per 1M tokens (+11%)
- Third Kimi K3 cached-price move in two weeks (Oct 6 cut, Oct 12 promo, Oct 14 hike)
- Together averages a pricing change every 7 days, 43 changes in the last year
- Standard Kimi K3 input ($3.00) and output ($15.00) per 1M tokens unchanged
THEIR RATIONALE
Together doesn't frame this one with a promo or announcement — it's a quiet table edit, but it fits a pattern of frequent fine-tuning on Kimi K3's cached rate specifically: cached input went from $0.27 to $0.30, two days after Together had just cut Kimi K3 token prices 10% on Oct 6 and days after adding a 10%-off Kimi K3 promo on Oct 12. Together is actively tuning this one model's economics rather than setting a price and leaving it.
THE SIGNAL
Together changes its pricing page roughly every 7 days on average, and it has touched Kimi K3 cached pricing three times in the last two weeks — a cut on Oct 6, a promo added Oct 12, and now this increase on Oct 14. Cached-token pricing is also where a lot of the market's attention is right now: 199 pricing changes across 152 companies in the last 180 days touched credits and usage billing in some form. For an inference provider, nudging the cached rate up a few cents is a lever that barely shows up to a casual reader but still directly affects margin on repeat-context workloads.
IN CONTEXT
Together has been unusually active on this one model's pricing: it cut Kimi K3 token prices 10% on Oct 6, added a 10%-off Kimi K3 promo two days before that lapsed on Oct 12, and now nudges the cached-input rate up to $0.30. That's three touches to Kimi K3 economics in two weeks, well inside Together's own cadence of a change roughly every 7 days (43 in the last year). It isn't alone in actively managing usage pricing right now — Elevenlabs pulled its Eleven v4 credits promo the same day, and Mintlify swapped free credits for a trial a day earlier.