DeepSeek V4.1 Flash 1017 vs Kimi K3.
Verdict by category, tale of the tape, shared benchmarks only, and cost at three workloads — all from the tracked data.
Verdict by category
| Category | DeepSeek V4.1 Flash 1017 | Kimi K3 | Winner |
|---|---|---|---|
| League Score | 73.7 | 74.6 | Kimi K3 |
| Coding (SWE-bench %) | — | — | — |
| Reasoning (GPQA %) | 90.9 | 93.5 | Kimi K3 |
| Context (tokens) | 1000000 | 1048576 | Kimi K3 |
| Speed (tok/s) | — | 44.8 | — |
| Price (blended $/1M) | $0.52 | $6.00 | DeepSeek V4.1 Flash 1017 (cheaper) |
Tale of the tape
| Spec | DeepSeek V4.1 Flash 1017 | Kimi K3 |
|---|---|---|
| Maker | DeepSeek | Moonshot AI |
| Released | Oct 17, 2026 | Jul 16, 2026 |
| League Score | 74 | 75 |
| Context | 1,000,000 tokens | 1,048,576 tokens |
| Max output | 384,000 | 131,072 |
| $/1M input | $0.30 | $3.00 |
| $/1M output | $1.20 | $15.00 |
| Weights | Open | Open |
| Speed | — | 44.8 t/s |
Head-to-head benchmarks (shared only)
| Benchmark | DeepSeek V4.1 Flash 1017 | Kimi K3 |
|---|---|---|
| LiveBench overall | 70.3 | 65.0 |
| GPQA Diamond | 90.9 | 93.5 |
Cost at three workloads
| Workload | DeepSeek V4.1 Flash 1017 | Kimi K3 |
|---|---|---|
| Light — 1M in / 250K out | $0.60 | $6.75 |
| Standard — 10M in / 2.5M out | $6.00 | $67.50 |
| Long-context — 50M in / 5M out | $21.00 | $225.00 |
The pick
Pick DeepSeek V4.1 Flash 1017 if you want the lower running cost.
Pick Kimi K3 if you want the higher overall League Score; you need the bigger context window.
All figures from the model database — sources and dates on each model page. Cost your own workload or compare a different pair.
Frequently asked questions
Which is better overall, DeepSeek V4.1 Flash 1017 or Kimi K3?
On overall League Score they are effectively tied — DeepSeek V4.1 Flash 1017 scores 74 and Kimi K3 scores 75. The category verdicts on this page are where the real differences live.
Which is cheaper, DeepSeek V4.1 Flash 1017 or Kimi K3?
DeepSeek V4.1 Flash 1017 is cheaper: $0.52 vs $6.00 blended per 1M tokens (3:1 input/output mix) — about 11.4x less per token.
Which has the bigger context window, DeepSeek V4.1 Flash 1017 or Kimi K3?
Kimi K3, with 1M of context against 1M for DeepSeek V4.1 Flash 1017.
Which is better at reasoning, DeepSeek V4.1 Flash 1017 or Kimi K3?
Kimi K3 leads on GPQA Diamond, 93.5% vs 90.9% on graduate-level science questions.
Are DeepSeek V4.1 Flash 1017 and Kimi K3 open source?
DeepSeek V4.1 Flash 1017 is open weights — it can be downloaded and self-hosted; Kimi K3 is open weights — it can be downloaded and self-hosted.