Kimi K3
Moonshot AI · Model release
Released Not recorded
Catalog checked 2026-09-28
- Input $/1M
- $3.00
- Output $/1M
- $15.00
- Context
- 1M
- Max output
- 1.05M
- Reasoning
- Yes
- Included in
- 13 plans
| Rate | Input | Output | Cache read |
|---|---|---|---|
| Standard | Input$3.00 | Output$15.00 | Cache read$0.30 |
Published rate checked 2026-09-28. Sources below.
Official platform lists cache writes at $3 per million tokens. Five-minute and one-hour storage are supported; duration-specific write pricing is not resolved here. No write duration is assumed.
Current published base rates. Context tiers, cache-write duration and other conditions may change the rate for a request. These are not a reconstructed historical invoice.
API token estimate
Explore a token scenario
Uncached text input and billed output at this exact model’s recorded Standard API list price. Totals across a scenario, without a claim about one-request capacity.
Standard API list price · USD / 1M tokens
Pricing checked Sep 28, 2026 · Rate record from Sep 28, 2026
Record dates describe catalog coverage; they do not establish provider price activation.
- Official standard token prices; checked September 28, 2026 checked Sep 28, 2026
Pricing record: kimi-k3-api-reference-20260928
- Context window
- 1,000,000
- Maximum output
- 1,048,576
- Input
- text · image · video
- Output
- text
- Reasoning
- Supported
- Tool calling
- Supported
- Structured output
- Supported
- Thinking
- Always enabled. Effort: low, high or max; max is the default.
- Output control
- max_completion_tokens defaults to 131,072 and can be set up to 1,048,576. This parameter ceiling does not add to the 1M context budget.
- Access
- Requires a successful top-up of at least $1. Cumulative top-ups determine API rate limits.
- Prompt caching
- Automatic prefix caching with 5-minute or 1-hour TTL. Partial cache blocks are billed as misses. Cache writes are charged separately.
- Tools
- Strict JSON Schema and tool calling are supported. Provider warns web search is being updated and is not recommended for production yet.
- Official specifications and conditions checked Sep 29, 2026
Benchmarks
Open benchmark sheet →Moonshot AI · Publication date unreported
Developer reported · Kimi K3 Tech Blog: Terminal-Bench 2.1
Coding
- Terminal-Bench 2.1
- 88.3%
Methodology & sources
Moonshot AI reports Kimi K3 evaluated with Kimi Code at max reasoning effort, temperature = 1.0 and top-p = 1.0.
These scores are reported by Moonshot AI. Model developers identify authorship; results may originate with another evaluator. Exact benchmark versions remain distinct. StackReplay calculates no composite score.
Terminal-Bench 2.1 · 88.3%
Multi-step tasks in a command-line environment.
Version: 2.1 · Reported benchmark score · percent · Higher is better · Task subset: Not separately specified
Moonshot AI · Publication date unreported · Developer reported
- Reporting source
- Moonshot AI · Kimi K3 Tech Blog: Terminal-Bench 2.1
- Evaluation origin
- Moonshot AI administered and reported the evaluation (source-computed)
- Effort
- max
- Harness
- Kimi Code (version unreported)
- Tools
- Not reported
- Fallback
- Not reported
- Provider
- Not reported
- Checked
- Oct 4, 2026
The raw table score is 88.3 in the Kimi K3 (max) column. The provider source omits the unit marker. StackReplay interprets the raw notation 88.3 on Terminal-Bench 2.1's published percent accuracy scale (https://www.tbench.ai/news/terminal-bench-2-1); numeric value is unchanged, with no arithmetic rescaling. The display % is this documented metric interpretation, not provider-explicit notation or leaderboard admission. Temperature = 1.0; top-p = 1.0. The original article publication date is unreported. Image asset dates, the weights-release statement, model release and October 4 check dates do not establish article publication.
Kimi Code version, trial count, scored task count and subset, suite revision beyond 2.1, evaluation start and completion dates, timeout, maximum output tokens, tools, exact endpoint/version, prompts, seeds and error handling are unreported.
$9.99 monthly; weekly and monthly usage windows follow the calendar.
Monthly cycle; purchased credits carry over without expiry.
New subscriptions use per-model allowances from September 28, 2026. Existing flat-$10 subscriptions change at their next renewal.
Monthly cycle; purchased credits carry over without expiry.
Monthly cycle; purchased credits carry over without expiry.
Monthly cycle; purchased credits carry over without expiry.
Credits reset on the first of the month; unused included credits do not carry over.
Credits reset on the first of the month; unused included credits do not carry over.
Credits reset on the first of the month; unused included credits do not carry over.
Credits reset on the first of the month; unused included credits do not carry over.
Credits reset on the first of the month; unused included credits do not carry over.
Monthly credit anniversary; included credits do not carry over.
Monthly subscription; one subscriber per workspace. Paid balance fallback is optional. Recurring charges continue until canceled through account settings. Reset anchors, rollover and a Go-specific refund policy are not established by the reviewed documentation.
Monthly subscription; one subscriber per workspace. Paid balance fallback is optional. Recurring charges continue until canceled through account settings. Reset anchors, rollover and a Go-specific refund policy are not established by the reviewed documentation.
Pricing, assumptions & evidence
Published base rate
Input $3.00 · Output $15.00 · Cache read $0.30 · Cache write Not listed
- Official standard token prices; checked September 28, 2026 checked Sep 28, 2026
- Official model specifications checked Sep 28, 2026
- Official specifications and conditions checked Sep 29, 2026
- GitHub pricing (official) checked Sep 23, 2026
- GitHub supported model and retirement tables; reviewed in the Sep 23 launch audit checked Sep 23, 2026
- Published token rates checked Sep 28, 2026
Missing token-category prices are not zero. Workload pricing applies exact recorded categories and admitted routes. No benchmark score or quality ranking is inferred from prices.
Aliases, routes and identity
Catalog ID kimi-k3
Model release (kind: release)
Developer: Moonshot AI
No additional exact aliases recorded.