#271 MoonshotAI: Kimi K3.1 Pro
low
fetch failed
- Cost
- $0.000
- Time
- 10.7s
- Tokens
- 0 tok
Moonshot AI
Release: 2026-10-24
Tested on: 2026-10-24 15:44
moonshotai/kimi-k3.1-pro::low
~2.8T total (~104B active)MoEClosedEstimated
Summary
Kimi K3.1 Pro scores 6.2 on AI BENCHY and ranks #271. It has 8.2 reliability, a 59.4% pass rate, $1.483 total cost, and 46.26s average response time.
What makes Kimi K3.1 Pro unique: It stands out most in Instructions following, where it ranks #1, while Puzzle Solving is its weakest area at #13.
Model facts
Researched on 2026-10-24
- Parameters
- ~2.8T total (~104B active)
- Architecture
- MoE
- Availability
- Closed
- License
- -
Best estimate from public evidence; the vendor did not disclose every value. API-only exact checkpoint on the research date. Moonshot says the full weights will be released by November 4; do not classify promised weights as available. Total and active parameter values are estimates inherited from the Kimi K3 backbone because the official Pro blog describes a post-trained K3 without restating exact counts. Native K3.1 Pro documentation requires thinking with low/high/max, which are the benchmarked modes. On the research date, the standard endpoint is $2/$10 per million; the same model has a faster $4/$20 endpoint. These benchmarks use standard speed.
6.2
Consistency
8.1
8.2
$1.483
Total Output Tokens
122,541
Total Input Tokens
128,710
Input Price
$2.000 / 1M
Output Price
$10.000 / 1M
Cache Read Price
$0.200 / 1M
Cache Write Price
N/A
Cache prices apply to input tokens. Reads reuse cached prompts; writes store them and can cost extra. Output tokens use the output price.
Flaky tests
5
Flaky tests had mixed outcomes across runs (at least one pass and one fail).
Hamster playing table tennis
Prompt: Create a detailed SVG illustration of a hamster playing table tennis.
Run history
| Tested on | Score | Reliability | Tests Correct | Total Cost | Compare |
|---|---|---|---|---|---|
| 2026-10-24 19:30 Re-test | 8.8 | 8.4 | $1.836 | Compare | |
| 2026-10-24 15:44 Initial run | 6.2 | 8.2 | $1.483 | Current run |
Run comparison
| Run | Benchmark coverage | Score | Consistency | Reliability | Tests Correct | Flaky tests | Total Output Tokens | Total Input Tokens | Total Cost | Response Time (avg) |
|---|---|---|---|---|---|---|---|---|---|---|
| 2026-10-24 15:44 · Initial run | 69/69 attempts | 6.2 | 8.1 | 8.2 | 11/23 | 5 | 122,541 | 128,710 | $1.483 | 46.26s |
| 2026-10-24 19:30 · Re-test | 69/69 attempts | 8.8 | 8.9 | 8.4 | 17/23 | 3 | 139,158 | 221,802 | $1.836 | 47.50s |
| Difference | — | -2.6 | -0.8 | -0.2 | -6 | +2 | -16617 | -93092 | -$0.353 | -1241ms |
Charts
Choose the first model, then click a second model to open a side-by-side page.
Top Models by Score
Score vs Total Cost
Response Time (avg)
Score vs Response Time (avg)
Total Output Tokens
Score vs Total Output Tokens
Quick Compare
Kimi K3.1 ProlowvsPareto 26.9defaultKimi K3.1 ProlowvsQwen3.8 MAXlowKimi K3.1 ProlowvsClaude Fable 5mediumKimi K3.1 ProlowvsGPT-5.4mediumKimi K3.1 ProlowvsMuse Spark 1.1mediumKimi K3.1 ProlowvsMuse Spark OpenmediumKimi K3.1 ProlowvsMuse Spark 1.2mediumKimi K3.1 ProlowvsDeepSeek V4.1 FlashhighKimi K3.1 ProlowvsMuse Spark 1.2highKimi K3.1 ProlowvsDeepSeek V4.1 Flashmedium
Category Breakdown
| Category | Score | Consistency | Tests Correct |
|---|---|---|---|
| Agentic | 3.5 | 8.3 | |
| Anti-AI Tricks | 8.7 | 7.9 | |
| Coding | 5.9 | 7.2 | |
| Combined | 6.5 | 10.0 | |
| Data parsing and extraction | 10.0 | 10.0 | |
| Domain specific | 3.5 | 4.4 | |
| General Intelligence | 10.0 | 10.0 | |
| Instructions following | 10.0 | 10.0 | |
| Puzzle Solving | 5.3 | 7.2 | |
| Tool Calling | 3.0 | 10.0 | |
| Trivia | 3.0 | 10.0 |