Kimi K3 Capability Score (Epoch Capabilities Index)

## What this measures

The current Epoch Capabilities Index (ECI) score of Kimi K3, an AI model from Moonshot AI. Epoch AI combines results from about 40 benchmarks (math, coding, science, reasoning, agentic tasks) into one scale, so models can be compared with each other and across years. One reading per month: the estimate Epoch published as of that month.

## Why it matters

Capability sets what a model can be trusted to do. Read next to the model's list price, it shows how much capability a dollar buys, and how fast that is changing across labs.

## How to read it

Higher is more capable. Compare it with other models' scores and with the frontier (ai.capability.eci_frontier). The scale has no natural ceiling and no direct translation into tasks or jobs.

## What it does not say

A benchmark score is not workplace performance: it misses reliability over long tasks, judgment and cost per task. Epoch re-estimates scores when it adds benchmarks, so a month's reading can differ from a later estimate of the same model.

## History

Single release: the model's current score from Epoch AI's latest release, one point a month from the month Palanor began tracking the model. The series is short by design, and there is no earlier history to backfill.

## Source

Epoch AI, Epoch Capabilities Index (epoch.ai/eci), eci_scores.csv in the benchmark_data.zip bundle. Licensed CC BY 4.0: "Epoch AI's data is free to use, distribute, and reproduce provided the source and authors are credited." Refreshed weekly by Palanor's epoch-sync job.