Skip to content
CostPerPrompt

Kimi K3.1 Pro (batch) API Pricing

Moonshot (Kimi) · context window 1M · prices updated 2026-11-21

Input per 1M tokens
$0.372
Output per 1M tokens
$5.58
Cached input per 1M
$0.1
73% cheaper than fresh input

Where Kimi K3.1 Pro (batch) sits on price

At $5.58 per million output tokens, Kimi K3.1 Pro (batch) is a premium tier model — 2.2× the median output price of $2.50, which puts it cheaper than 34% and pricier than 66% of the models we track. It sits 7th cheapest of the 10 Moonshot (Kimi) models we track. Output costs 15.0× more than input here, so anything that makes the model write less — tighter instructions, structured output, lower max_tokens — moves the bill more than trimming the prompt.

Output tokens per $1
179,115
One full context fill
$0.3901
Cheaper than
34% of tracked models

Cached input is 73% cheaper. On an input-heavy workload you need roughly a 34% cache-hit rate to take 25% off the input line — reachable for chatbots and agents that resend the same system prompt and history.

What real workloads cost on Kimi K3.1 Pro (batch)

These three workloads are the ones teams actually run on a premium-priced model — a $5.58/1M model is not bought for the same job as one ten times the price.

Workload Per request Per month
Support chatbot — 500 conversations/day (5K in / 1.4K out) Conversation history re-sent each turn — the classic prompt-caching win. $0.0097 $145
Coding agent — 300 sessions/day (60K in / 12K out) Multi-step loops re-read the same files; caching matters more than raw price. $0.0893 $804
Long-context analysis — 200 runs/day (250K in / 4K out) Only possible on large-context models; input cost is nearly the whole bill. $0.1153 $692

Model your exact traffic in the API cost calculator — it preloads Kimi K3.1 Pro (batch) with caching and batch options.

Kimi K3.1 Pro (batch) price history

Moonshot (Kimi) last cut Kimi K3.1 Pro (batch)'s output price on Nov 17, 2026 (-7% per output token), from $0.7/$6.00 to $0.372/$5.58 per 1M in/out. That is one of 3 repricings in the 9 days we have tracked it — worth knowing before you hard-code today's rate into a budget.

Date Input /1M Output /1M Output change
Nov 12, 2026 (tracking began) $0.7 $6.00 —
Nov 15, 2026 $0.275 $4.13 -31%
Nov 16, 2026 $0.7 $6.00 +45%
Nov 17, 2026 $0.372 $5.58 -7%

Change-points from our daily price snapshots (tracking since Nov 12, 2026; intraday moves between snapshots are not captured).

Kimi K3.1 Pro (batch) vs GLM 5.3 Prime

The closest-priced alternative from another vendor is GLM 5.3 Prime (Z.ai (GLM)) — 5% cheaper on output, with $1.31 more per million input tokens. When two models land this close on price, the decision is quality on your own workload, not the price sheet: run 50 real requests through both and compare.

See GLM 5.3 Prime pricing →

Cheaper alternatives

More Moonshot (Kimi) models

Frequently asked questions

› How much does the Kimi K3.1 Pro (batch) API cost?

Kimi K3.1 Pro (batch) costs $0.372 per million input tokens and $5.58 per million output tokens, with cached input at $0.1 per million (73% cheaper). That works out to roughly 179,115 output tokens per dollar.

› What does the support chatbot workload cost on Kimi K3.1 Pro (batch)?

Support chatbot — 500 conversations/day (5K in / 1.4K out) costs about $0.0097 per request and $145 per month on Kimi K3.1 Pro (batch). Conversation history re-sent each turn — the classic prompt-caching win.

› What does it cost to fill Kimi K3.1 Pro (batch)'s 1M context window?

Sending 1M of input in a single request costs $0.3901 at $0.372 per million tokens — before any output. With prompt caching that same fill drops to about $0.1049 on repeat requests. This is why large context windows are cheap to advertise and expensive to actually use.

› Is Kimi K3.1 Pro (batch) worth the price?

Kimi K3.1 Pro (batch) sits in the middle of the market (7 of 10 by price within Moonshot (Kimi)). The honest test is a routing experiment: send the same 200 real requests to this model and to a tier below, and compare failure rate against the price gap — most teams find a majority of traffic never needed the pricier model.