Buying guide
| Input | $0.300 | — |
| Output | $1.20 | — |
| Cached input | $0.0060 | — |
| Context window | 1,000K | — |
| Blended /1M | $0.750 | — |
You need the highest benchmark scores and can absorb the higher per-token price.
You are optimizing for cost at scale and your workloads tolerate the score gap.
DeepSeek V4.1-Flash 0910 leads on 13 of 15 shared benchmarks.
| Dimension | DeepSeek V4.1-Flash 0910 | Kimi K3 |
|---|---|---|
| GPQA Diamond | 90.9% | 92.9% |
| HLE | 36.8% | 43.5% |
| Codeforces (Rating) | 3471 | — |
| MathArena Apex | 65.6% | 65.6% |
| Terminal-Bench 2.1 | 90.6% | 88.3% |
| Terminal-Bench 3.0 | 30% | 17.7% |
| Terminal-Bench 4.0 | 31.2% | 12.6% |
| DeepSWE v1.1 | 74.2% | 67.5% |
| ProgramBench | 20.3% | 17.5% |
| NL2Repo-Bench | 65.4% | 58% |
| CyberGym | 88.1% | 80% |
| SEC-Bench Pro | 62.8% | — |
| ExploitGym | 15.3% | — |
| HLE (w/tools) | 63.9% | 59.8% |
| Automation-Bench | 54.8% | 46.7% |
| Agents' Last Exam | 31.8% | 27.6% |
| Chartography (w/tools) | 78.9% | 68.1% |
| BabyVision (w/tools) | 89.6% | 85.7% |
| ZeroBench-main (w/tools) | 49% | 41% |
| Blended price | $0.750 /1M | — |
| Full priceInput · Output · Cached input | $0.300 · $1.20 · $0.0060 | — |
| Context window | 1,000K | — |
Adjust the monthly volume — every model recomputes instantly.
Estimates use list prices and ignore cached input, batch discounts, and off-peak rates.