DeepSeek vs Kimi
DeepSeek's DeepSeek and Moonshot AI's Kimi, compared through their current models on the Respan Index. The strongest DeepSeek model today is DeepSeek-V4.1-Flash-1017 (#23 on the Respan Index, 59.3); the strongest Kimi model is Kimi K3.1 Pro (#9, 71.4). DeepSeek's best models lead in agents and reasoning; Kimi's lead in coding, knowledge and human preference.
Flagship vs flagship
The highest ranked DeepSeek and Kimi models on the Respan Index.
| Spec | DeepSeek-V4.1-Flash-1017 | Kimi K3.1 Pro |
|---|---|---|
| Respan Index | 59.3 (#23) | 71.4 (#9) |
| Blended price per 1M | $0.1 | $1.94 |
| Context window | 1M | 1.0M |
| Input / output per 1M | $0.036 / $0.294 | $0.253 / $7 |
Best in each category
Each family's top scorer per category, scored 0 to 100, and how far ahead the winner is. Leads under a point are close to even.
| Category | DeepSeek | Kimi |
|---|---|---|
| Coding | 48.1DeepSeek-V4.1-Flash-1017 | 68.7Kimi K3.1 Pro |
| Agents | 76.3DeepSeek-V4.1-Flash-1017 | 70.3Kimi K3.1 Pro |
| Reasoning | 61.1DeepSeek-V4.1-Flash-1017 | 54.8Kimi K3 |
| Knowledge & language | 52.6DeepSeek V4 Pro | 52.9Kimi K3 |
| Human preference | 57.5DeepSeek V4.1 Flash | 58.3Kimi K3.1 Pro |
Both lineups, ranked
The top 6 DeepSeek and Kimi models in Respan Index order, with blended list price per 1M tokens.
| Rank | Model | Index | Price |
|---|---|---|---|
| #9 | Kimi K3.1 Pro | 71.4 | $1.94/M |
| #23 | DeepSeek-V4.1-Flash-1017 | 59.3 | $0.1/M |
| #35 | Kimi K3 | 52.8 | $2.52/M |
| #37 | DeepSeek V4.1 Flash | 52.5 | $0.089/M |
| #45 | DeepSeek V4 Pro | 49.2 | $0.585/M |
| #56 | DeepSeek V4 Flash | 45.3 | $0.113/M |
| #77 | Kimi K2.6 | 36.3 | $0.742/M |
| #78 | Kimi K2.7 Code | 36.1 | $1.36/M |
| #83 | DeepSeek-V4-Flash (Preview) | 32.3 | $0.056/M |
| #96 | Kimi K2.5 | 29.3 | $0.9/M |
| #104 | DeepSeek V3.2 | 26.6 | $0.29/M |
| #146 | Kimi K2 | 17.6 | $1/M |
The cheapest ranked DeepSeek model is DeepSeek-V4-Flash (Preview) at $0.056 per 1M tokens blended; the cheapest Kimi model is Kimi K2.6 at $0.742.
FAQ
Is DeepSeek better than Kimi?
The strongest DeepSeek model today is DeepSeek-V4.1-Flash-1017 (#23 on the Respan Index, 59.3); the strongest Kimi model is Kimi K3.1 Pro (#9, 71.4). DeepSeek's best models lead in agents and reasoning; Kimi's lead in coding, knowledge and human preference.
Is DeepSeek or Kimi better for coding?
On the Respan Index coding score, DeepSeek's best is DeepSeek-V4.1-Flash-1017 at 48.1 and Kimi's best is Kimi K3.1 Pro at 68.7. The coding score blends SWE-bench, Terminal-Bench and other coding benchmarks.
Which is cheaper, DeepSeek or Kimi?
The cheapest ranked DeepSeek model is DeepSeek-V4-Flash (Preview) at $0.056 per 1M tokens blended; the cheapest Kimi model is Kimi K2.6 at $0.742.
What are the best DeepSeek and Kimi models right now?
DeepSeek-V4.1-Flash-1017 for DeepSeek and Kimi K3.1 Pro for Kimi, ranked #23 and #9 on the Respan Index, which blends public benchmarks across coding, agents, reasoning, knowledge and human preference.