Kimi K3 vs DeepSeek V4.1 Flash 1017
Kimi K3 is from Moonshot. DeepSeek V4.1 Flash 1017 is from DeepSeek. Kimi K3 costs $15.00 per million output tokens. DeepSeek V4.1 Flash 1017 costs $0.600. DeepSeek V4.1 Flash 1017 leads on the Epoch Capabilities Index: Kimi K3 scores 158.3 and DeepSeek V4.1 Flash 1017 scores 158.8. A typical request, about 1,000 words in and 400 words out, costs $0.0077 on Kimi K3 and $0.0005 on DeepSeek V4.1 Flash 1017.
Prices updated . Benchmarks updated .
| Kimi K3 | DeepSeek V4.1 Flash 1017 | |
|---|---|---|
| Lab | Moonshot | DeepSeek |
| You send / 1M | $0.192 | $0.150 |
| It replies / 1M | $15.00 | $0.600 |
| Context | 1.05M | 1.05M |
| Launched | 2026-07-16 | 2026-10-17 |
| Weights | Open weights, restricted | Open weights |
| Typical request | $0.0077 | $0.0005 |
| ECI | 158.3 | 158.8 |
| GPQA | 93.1% | 93.2% |
| ARC-AGI-2 | 60.4% | 78.1% |
| HLE | n/a | n/a |
| AIME | 97.2% | 99.4% |
| SimpleQA | 50.6% | 42.5% |
| CritPt | 23.4% | 24.0% |
| Arena | 1,476 | n/a |
| Arena web | 1,656 | n/a |
Which is cheaper?
DeepSeek V4.1 Flash 1017 costs less to run: $0.600 per million output tokens, against $15.00 for Kimi K3. Input tokens are $0.192 per million for Kimi K3 and $0.150 per million for DeepSeek V4.1 Flash 1017. A typical request, about 1,000 words in and 400 words out, costs $0.0077 on Kimi K3 and $0.0005 on DeepSeek V4.1 Flash 1017.
Which scores higher?
DeepSeek V4.1 Flash 1017 scores higher on Epoch Capabilities Index: 158.8 against 158.3 for Kimi K3. That index is an independent composite from Epoch AI. Lab-reported scores are kept off this comparison. Context windows are 1.05M and 1.05M. Prices are US dollars per million tokens at current public rates.
Verdict
DeepSeek V4.1 Flash 1017 is both cheaper and higher on Epoch Capabilities Index than Kimi K3, at $0.600 per million output tokens and a score of 158.8. Context is 1.05M for Kimi K3 and 1.05M for DeepSeek V4.1 Flash 1017. Use the model pages for every score on file, including lab-reported numbers.
FAQ
Which is cheaper, Kimi K3 vs DeepSeek V4.1 Flash 1017?
DeepSeek V4.1 Flash 1017 costs less to run: $0.600 per million output tokens, against $15.00 for Kimi K3. Input tokens are $0.192 per million for Kimi K3 and $0.150 per million for DeepSeek V4.1 Flash 1017. A typical request, about 1,000 words in and 400 words out, costs $0.0077 on Kimi K3 and $0.0005 on DeepSeek V4.1 Flash 1017.
Which scores higher, Kimi K3 vs DeepSeek V4.1 Flash 1017?
DeepSeek V4.1 Flash 1017 scores higher on Epoch Capabilities Index: 158.8 against 158.3 for Kimi K3. That index is an independent composite from Epoch AI. Lab-reported scores are kept off this comparison. Context windows are 1.05M and 1.05M. Prices are US dollars per million tokens at current public rates.
What is the verdict for Kimi K3 vs DeepSeek V4.1 Flash 1017?
DeepSeek V4.1 Flash 1017 is both cheaper and higher on Epoch Capabilities Index than Kimi K3, at $0.600 per million output tokens and a score of 158.8. Context is 1.05M for Kimi K3 and 1.05M for DeepSeek V4.1 Flash 1017. Use the model pages for every score on file, including lab-reported numbers.
Open this pair in the comparison view
Sources and licenses
- OpenRouter: Prices, context windows, modalities and the model list come from the OpenRouter public API. Data: OpenRouter.
- Vercel AI Gateway: The no-training flag comes from the Vercel AI Gateway model list.
- models.dev: Lab launch dates, open-weights flags and knowledge cutoffs come from models.dev. Copyright (c) 2025 models.dev, MIT License.
- Epoch AI: Independent benchmark scores and the Epoch Capabilities Index (ECI) come from Epoch AI's Benchmarking Hub, used under CC BY 4.0. Epoch AI, Capabilities & benchmarking, epoch.ai/benchmarks. Scores Epoch marks as external keep their original license, and the source is named on each score.
- LMArena: Arena Elo (text, webdev and vision) comes from the LMArena leaderboard dataset (lmarena-ai/leaderboard-dataset), CC BY 4.0, leaderboard date 2026-10-28.
- Lab-reported scores, when an editor has added them, are the lab's own numbers. Each one links to the source and is labeled lab-reported. They are not used to rank the lists.
- What a score measures: the benchmark guide.