Kimi K2.7 Code vs Kimi K3 (max)
Which is better for the money: Kimi K2.7 Code by Kimi or Kimi K3 (max) by Kimi? Intelligence, total benchmark cost, and end-to-end time from Artificial Analysis & other sources, snapshot 2026-08-07.
Trade-off
Verdict
Depends on your budget
Kimi K3 (max) scores 16.7 points higher (59.7 vs 43.0) but costs 4.5× as much to run the benchmark ($2425 vs $545; flagship vs premium cost classes). They are not alternatives at one budget: pick Kimi K3 (max) when you need intelligence above 43.0, and Kimi K2.7 Code when 43.0 is enough.
| Metric | Kimi K2.7 Code | Kimi K3 (max) | Difference |
|---|---|---|---|
| Intelligence (higher is better) | 43.0 | 59.7 | +38.8% Kimi K3 (max) is 38.8% more intelligent |
| Total AA cost (lower is better) | $544.58 | $2,425.11 | +345.3% Kimi K3 (max) is 4.5× more expensive |
| End-to-end time (lower is better) | 66.82s | 67.69s | +1.3% Kimi K3 (max) is 1.3% slower |
| Price per 1M tokens (in / out) | $0.95 / $4.00 | $3.00 / $15.0 | |
| Input modalities | Not reported | Not reported |
Kimi K3 (max) vs Kimi K2.7 Code: 38.8% more intelligence, 345.3% more cost to run the AA benchmark, and 1.3% slower end-to-end.
Today's Pareto frontier
Against the frontier
- Kimi K2.7 Code is off the frontier: at the same or lower cost, GPT-5.6 Luna (max) scores 52.30 for $172.
- Kimi K3 (max) is off the frontier: at the same or lower cost, Claude Opus 5 (Adaptive Reasoning, High Effort) scores 61.50 for $1974.
Kimi K2.7 CodeKimi K3 (max)Today's Pareto frontierEvery other model
Matchup
Compare something else
Agents:
Accept: text/markdown or ?format=md returns this page as Markdown. JSON: /api/v1/compare/kimi-k2-7-code/kimi-k3 · Markdown: /api/v1/compare/kimi-k2-7-code/kimi-k3?format=md · llms.txt · agent skill