DeepSeek V3 (Dec '24) vs Kimi K2.7 Code
Which is better for the money: DeepSeek V3 (Dec '24) by DeepSeek or Kimi K2.7 Code by Kimi? Intelligence, total benchmark cost, and end-to-end time from Artificial Analysis & other sources, snapshot 2026-08-19.
Trade-off
Verdict
Depends on your budget
Kimi K2.7 Code scores 28.8 points higher (43.0 vs 14.2) but costs 27× as much to run the benchmark ($545 vs $20.0; premium vs budget cost classes). They are not alternatives at one budget: pick Kimi K2.7 Code when you need intelligence above 14.2, and DeepSeek V3 (Dec '24) when 14.2 is enough.
| Metric | DeepSeek V3 (Dec '24) | Kimi K2.7 Code | Difference |
|---|---|---|---|
| Intelligence (higher is better) | 14.2 | 43.0 | +202.8% Kimi K2.7 Code is 3.0× more intelligent |
| Total AA cost (lower is better) | $20.02 | $544.58 | +2620.2% Kimi K2.7 Code is 27× more expensive |
| End-to-end time (lower is better) | Not reported | 75.91s | — not comparable |
| Price per 1M tokens (in / out) | $0.36 / $0.89 | $0.95 / $4.00 | |
| Input modalities | Not reported | Not reported |
Kimi K2.7 Code vs DeepSeek V3 (Dec '24): 202.8% more intelligence, 2620.2% more cost to run the AA benchmark, and end-to-end time not fully reported.
Today's Pareto frontier
Against the frontier
- DeepSeek V3 (Dec '24) is off the frontier: at the same or lower cost, GPT-5.6 Luna (low) scores 33.90 for $13.7.
- Kimi K2.7 Code is off the frontier: at the same or lower cost, Gemini 3.7 Flash (high) scores 56.00 for $485.
DeepSeek V3 (Dec '24)Kimi K2.7 CodeToday's Pareto frontierOpen-weights frontierEvery other model
Matchup
Compare something else
Agents:
Accept: text/markdown or ?format=md returns this page as Markdown. JSON: /api/v1/compare/deepseek-v3/kimi-k2-7-code · Markdown: /api/v1/compare/deepseek-v3/kimi-k2-7-code?format=md · llms.txt · agent skill