Model comparison
DeepSeek V4 Flash (Max) vs Kimi K3
Head-to-head evidence from 25 shared benchmark results across 5 categories. Overall scores shown here use the public BenchAlign v5 ranking lane.
Public leaderboard positions: DeepSeek V4 Flash (Max) unranked (Not scored); Kimi K3 #4 (Supported). Intervals and evidence labels describe ranking uncertainty, not a guarantee for a specific workload.
Evidence parity. DeepSeek V4 Flash (Max) and Kimi K3 share 25 comparable benchmark results. 2 of 8 categories are comparable. 20 results are unique to DeepSeek V4 Flash (Max); 32 to Kimi K3.
Updated July 23, 2026- Shared results
- 25
- DeepSeek V4 Flash (Max) only
- 20
- Kimi K3 only
- 32
- Comparable categories
- 2 / 8
Treat this as a split decision. DeepSeek V4 Flash (Max) makes more sense if you want the cheaper token bill; Kimi K3 is the better fit if agentic is the priority or you need the larger 1.05M context window.
Confidence note. This is a partial-evidence comparison with 25 shared benchmark results across 5 evidence categories; 2 of 8 categories currently have scoreable aggregates for both models. Treat the verdict as directional until coverage is more balanced.
Why this result
DeepSeek V4 Flash (Max) and Kimi K3 finish on the same BenchAlign overall score, so this is less about a single winner and more about where the edge shows up. The BenchAlign headline says tie; the benchmark table is where the real choice happens.
Kimi K3 is also the more expensive model on tokens at $3.00 input / $15.00 output per 1M tokens, versus $0.14 input / $0.28 output per 1M tokens for DeepSeek V4 Flash (Max). That is roughly 53.6x on output cost alone. Kimi K3 gives you the larger context window at 1.05M, compared with 1M for DeepSeek V4 Flash (Max).
Category breakdown
Exact category averages are shown below. Not measured means BenchLM does not have enough sourced public coverage for that model and category.
| Category | DeepSeek V4 Flash (Max) | Δ | Kimi K3 |
|---|---|---|---|
| Agentic | DeepSeek V4 Flash (Max)63.8 | Margin→ 25.7 | Kimi K389.5 |
| Knowledge | DeepSeek V4 Flash (Max)55.3 | Margin→ 5.7 | Kimi K361.0 |
| Coding | DeepSeek V4 Flash (Max)68.8 | MarginNo overlap | Kimi K3Not measured |
| Math | DeepSeek V4 Flash (Max)94.8 | MarginNo overlap | Kimi K3Not measured |
| Multimodal | DeepSeek V4 Flash (Max)Not measured | MarginNo overlap | Kimi K378.5 |
Decisive benchmark drivers
The largest measured benchmark gaps in this matchup, with exact reported values.
More
- Source ↗
Terminal-Bench 2.0
AgenticA 56.9%B 88.3%Winner: Kimi K3Δ 31.4Terminal-Bench 2.0: DeepSeek V4 Flash (Max) scored 56.9%; Kimi K3 scored 88.3%. Kimi K3 wins this benchmark. - Source ↗
HLE
KnowledgeA 34.8%B 56%Winner: Kimi K3Δ 21.2HLE: DeepSeek V4 Flash (Max) scored 34.8%; Kimi K3 scored 56%. Kimi K3 wins this benchmark. - Source ↗
BrowseComp
AgenticA 73.2%B 91.2%Winner: Kimi K3Δ 18BrowseComp: DeepSeek V4 Flash (Max) scored 73.2%; Kimi K3 scored 91.2%. Kimi K3 wins this benchmark. - Source ↗
GPQA
KnowledgeA 88.1%B 93.5%Winner: Kimi K3Δ 5.4GPQA: DeepSeek V4 Flash (Max) scored 88.1%; Kimi K3 scored 93.5%. Kimi K3 wins this benchmark.
Operational comparison
Runtime and commercial metrics are compared only when both models have a complete sourced value.
| Metric | DeepSeek V4 Flash (Max) | Kimi K3 | Comparison |
|---|---|---|---|
| Input / output priceUSD per 1M tokens | DeepSeek V4 Flash (Max)$0.14 input / $0.28 output | Kimi K3$3 input / $15 output | DeepSeek V4 Flash (Max) has the lower combined listed price. |
| Generation speedtokens per second | DeepSeek V4 Flash (Max)Not available | Kimi K3Not available | A complete speed comparison is not available. |
| First-answer latencyseconds to first token | DeepSeek V4 Flash (Max)Not available | Kimi K3Not available | A complete latency comparison is not available. |
| Context windowmaximum listed tokens | DeepSeek V4 Flash (Max)1M | Kimi K31.05M | Kimi K3 lists the larger context window. |
Benchmark Deep Dive
AgenticKimi K3 wins25 benchmarks
| Benchmark | DeepSeek V4 Flash (Max) | Kimi K3 | Result |
|---|---|---|---|
| Terminal-Bench 2.0Source | 56.9% | 88.3% | Kimi K3 leads |
| BrowseCompSource | 73.2% | 91.2% | Kimi K3 leads |
| HLE w/ toolsSource | 45.1% | — | Not comparable |
| MCP AtlasSource | 69% | 84.2% | Kimi K3 leads |
| GDPval-AASource | 1189 | 1679 | Kimi K3 leads |
| ToolathlonSource | 47.8% | — | Not comparable |
| AA Agentic IndexSource | 31.1% | 50.1% | Kimi K3 leads |
| τ²-bench resultsSource | 95% | — | Not comparable |
| GDPval-AASource | 34.4% | 59.0% | Kimi K3 leads |
| AA BriefcaseSource | 831 | 1543 | Kimi K3 leads |
| AA EnterpriseOps-GymSource | 39.6% | 45.3% | Kimi K3 leads |
| AA Harvey LABSource | 81.3% | 94.6% | Kimi K3 leads |
| AA ITBenchSource | 31.5% | 47.7% | Kimi K3 leads |
| AA Tau3 BankingSource | 22.9% | 33.4% | Kimi K3 leads |
| terminalBenchHardSource | 35.6% | — | Not comparable |
| DeepSearchQASource | — | 95.0% | Not comparable |
| Toolathlon-VerifiedSource | — | 73.2% | Not comparable |
| AutomationBenchSource | — | 30.8% | Not comparable |
| JobBenchSource | — | 52.9% | Not comparable |
| APEX-AgentsSource | — | 37.6% | Not comparable |
| SpreadsheetBench 2Source | — | 34.8% | Not comparable |
| DECK-BenchSource | — | 73.5% | Not comparable |
| AA AutomationBenchSource | — | 52.7% | Not comparable |
| aaTerminalBench21Source | — | 85% | Not comparable |
| APEX-Agents-AASource | — | 41.3% | Not comparable |
Coding14 benchmarks
| Benchmark | DeepSeek V4 Flash (Max) | Kimi K3 | Result |
|---|---|---|---|
| CodeforcesSource | 3052.0 | — | Not comparable |
| SWE-bench VerifiedSource | 79% | — | Not comparable |
| SWE-bench ProSource | 52.6% | — | Not comparable |
| SWE MultilingualSource | 73.3% | — | Not comparable |
| Terminal-Bench 2.0Source | 56.9% | — | Not comparable |
| AA Coding IndexSource | 56.2% | 76.2% | Kimi K3 leads |
| AA-SciCodeSource | 44.9% | 58.7% | Kimi K3 leads |
| deepSweSource | — | 67.5% | Not comparable |
| FrontierSWESource | — | 81.2% | Not comparable |
| ProgramBenchSource | — | 77.8% | Not comparable |
| Kimi Code Bench v2Source | — | 72.9% | Not comparable |
| sweMarathonSource | — | 42% | Not comparable |
| PostTrain BenchSource | — | 36.6% | Not comparable |
| MLS-Bench LiteSource | — | 48.3% | Not comparable |
Reasoning4 benchmarks
KnowledgeKimi K3 wins14 benchmarks
| Benchmark | DeepSeek V4 Flash (Max) | Kimi K3 | Result |
|---|---|---|---|
| MMLU-ProSource | 86.2% | — | Not comparable |
| SimpleQASource | 34.1% | — | Not comparable |
| Chinese-SimpleQASource | 78.9% | — | Not comparable |
| GPQASource | 88.1% | 93.5% | Kimi K3 leads |
| GPQA-DSource | 88.1% | 93.5% | Kimi K3 leads |
| HLESource | 34.8% | 56% | Kimi K3 leads |
| Artificial Analysis Intelligence IndexSource | 40.3% | 57.1% | Kimi K3 leads |
| AA-GPQA DiamondSource | 89.4% | 93.5% | Kimi K3 leads |
| AA-HLESource | 32.1% | 44.3% | Kimi K3 leads |
| AA-Omniscience IndexSource | -22.9% | 18.4% | Kimi K3 leads |
| AA-Omniscience AccuracySource | 37.2% | 46.0% | Kimi K3 leads |
| AA-Omniscience Hallucination RateSource | 95.8% | 50.9% | Kimi K3 leads |
| AA Openness IndexSource | 50.0% | — | Not comparable |
| HLE w/o toolsSource | — | 43.5% | Not comparable |
Math4 benchmarks
Multimodal15 benchmarks
| Benchmark | DeepSeek V4 Flash (Max) | Kimi K3 | Result |
|---|---|---|---|
| Design Arena WebsiteSource | 1238 | 1386 | Kimi K3 leads |
| OfficeQA ProSource | — | 63.3% | Not comparable |
| MMMU-ProSource | — | 81.6% | Not comparable |
| MMMU-Pro w/ PythonSource | — | 83.4% | Not comparable |
| CharXiv w/o toolsSource | — | 84.8% | Not comparable |
| CharXivSource | — | 91.3% | Not comparable |
| MathVisionSource | — | 94.3% | Not comparable |
| MathVision w/ PythonSource | — | 97.8% | Not comparable |
| BabyVision w/ PythonSource | — | 85.7% | Not comparable |
| ZeroBenchSource | — | 23.0% | Not comparable |
| ZeroBench w/ PythonSource | — | 41.0% | Not comparable |
| WorldVQA ForceAnswerSource | — | 51.0% | Not comparable |
| OmniDocBenchSource | — | 91.1% | Not comparable |
| PerceptionBenchSource | — | 58.5% | Not comparable |
| AA-MMMU-ProSource | — | 80.5% | Not comparable |
Inst. Following1 benchmarks
| Benchmark | DeepSeek V4 Flash (Max) | Kimi K3 | Result |
|---|---|---|---|
| AA-IFBenchSource | 79.2% | — | Not comparable |
Frequently Asked Questions (3)
Which is better, DeepSeek V4 Flash (Max) or Kimi K3?
DeepSeek V4 Flash (Max) and Kimi K3 are tied on the BenchAlign overall score, so the right pick depends on which category matters most for your use case.
Which is better for knowledge tasks, DeepSeek V4 Flash (Max) or Kimi K3?
Kimi K3 has the edge for knowledge tasks in this comparison, averaging 61 versus 55.3. Inside this category, AA-Omniscience Hallucination Rate is the benchmark that creates the most daylight between them.
Which is better for agentic tasks, DeepSeek V4 Flash (Max) or Kimi K3?
Kimi K3 has the edge for agentic tasks in this comparison, averaging 89.5 versus 63.8. Inside this category, AA Briefcase is the benchmark that creates the most daylight between them.
Related Comparisons
Explore More
Choose a model with this week’s evidence
Join 2,000+ readers for ranking moves, pricing changes, and the claims that still need proof.
One email each week. Unsubscribe anytime.