DeepSeekDeepSeek V4 Flash 0420
DeepSeek V4 Flash 0420 is a DeepSeek open-weights reasoning model, released 24 Apr 2026. Its best configuration (max reasoning effort) ranks #191 of 784 on the BenchLeader Index at 57.0 ±6.0, in the upper half. The ± is the point: 301 other configurations score within that range, so they and this one cannot be told apart on quality alone — price and speed are what separate them. It scores highest in instruction following (77) and lowest in knowledge (51). At $0.140 per million tokens blended it is among the cheapest fifth of ranked models. Output speed of 47 tokens per second puts it in the slowest quarter, with a first answer in 120.8 s. It has been measured at 3 reasoning-effort settings; this summary describes the best-scoring one, and the tabs above switch between them. Last measured 24 Apr 2026.stale: no new result in six months
- Blended price
- $0.140/M
- $0.110 in · $0.230 out
- Output speed
- 47 tok/s
- measured by Artificial Analysis
- First answer
- 121 s
- Context
- 1M
- Full answer
- 131 s
- median, reasoning included
- Cost per run
- $0.090
- one full Intelligence Index run
- Released
- 24 Apr 2026
- Overall index57
- Reasoning53
- Agents & tools61
- Knowledge51
- Instruction following77
- Long context62
- Composite57
Versions
DeepSeek has shipped 2 models under this name. Each is ranked on its own results; a newer version often has fewer results so far, which holds its index nearer the average until more arrive.
| Model | Released | Index | Rank |
|---|---|---|---|
| DeepSeek V4 Flashmax | 31 Jul 2026 | 59.4 | #129 |
| DeepSeek V4 Flash 0420maxthis page | 24 Apr 2026 | 57.0 | #191 |
Reasoning-effort configurations
The same model behaves differently depending on how much it is allowed to think. Each row is one setting, scored only on the benchmarks that were run at that setting. “Not stated” collects results from publishers that did not say which setting they used; for a reasoning model that is usually its thinking mode, but we do not assume it. Pick a setting here or at the top of the page to see its price, speed and category scores.
| Effort | Index | Rank | Speed | First answer | Chat reply cost | Categories |
|---|---|---|---|---|---|---|
| no reasoning | 49.0 | #410 | – | – | $0.0001 | Agents & tools 64 · Composite 51 · Instruction following 49 · Knowledge 42 · Long context 45 · Reasoning 41 |
| high | 55.4 | #232 | – | – | $0.0001 | Agents & tools 59 · Composite 57 · Instruction following 72 · Knowledge 51 · Long context 61 · Reasoning 47 |
| maxbest | 57.0 | #191 | 47 tok/s | 121 s | $0.0001 | Agents & tools 61 · Composite 57 · Instruction following 77 · Knowledge 51 · Long context 62 · Reasoning 53 |
Benchmark results
See all LLM benchmarks. One column per reasoning effort. Rank is among every configuration of every model on that benchmark. Hover a score for the run it came from.
Reasoning
| Benchmark | no reasoning | high | max | Source | Trend |
|---|---|---|---|---|---|
| GPQA Diamond (AA)not in index | 71.6%#285 | 86.7%#107 | 89.4%#78 | Artificial Analysis | |
| Humanity's Last Exam (AA)not in index | 7.8%#356 | 30.3%#147 | 34.9%#123 | Artificial Analysis | |
| CritPt | 0.3%#270 | 3.4%#174 | 7.1%#143 | Artificial Analysis |
Coding
| Benchmark | high | max | Source | Trend |
|---|---|---|---|---|
| SciCode (AA)not in index | 40.2%#171 | 45.3%#144 | Artificial Analysis | |
| Terminal-Bench 4.0 (AA)not in index | 3.0%#107 | 2.5%#112 | Artificial Analysis | |
| Terminal-Bench 2.1 (AA)not in index | 56.9%#103 | 61.8%#92 | Artificial Analysis |
Agents & tools
| Benchmark | no reasoning | high | max | Source | Trend |
|---|---|---|---|---|---|
| Terminal-Bench Hard | 34.1%#89 | 38.6%#60 | 35.6%#74 | Artificial Analysis | |
| τ²-Bench Telecom (AA)not in index | 94.4%#30 | 95.6%#19 | 95.0%#26 | Artificial Analysis | |
| GDPval-AA v2.1not in index | – | 24.9%#163 | 26.7%#155 | Artificial Analysis | |
| τ³-Banking (AA) | – | 26.2%#78 | 30.9%#64 | Artificial Analysis | |
| ITBench SRE (AA)not in index | – | – | 31.5%#31 | Artificial Analysis | |
| Analyst Agent (AA)not in index | – | – | 25.0%#32 | Artificial Analysis | |
| AutomationBenchnot in index | – | 28.5%#118 | 28.0%#121 | Artificial Analysis | |
| GDP.pdf (AA)not in index | – | 10.4%#127 | 10.8%#126 | Artificial Analysis | |
| EnterpriseOps-Gymnot in index | – | – | 39.6%#30 | Artificial Analysis |
Knowledge
| Benchmark | no reasoning | high | max | Source | Trend |
|---|---|---|---|---|---|
| AA-Omniscience | -44.0#342 | -23.1#254 | -23.9#258 | Artificial Analysis | |
| AA-Omniscience: accuracynot in index | 26.0%#220 | 34.9%#153 | 36.8%#145 | Artificial Analysis | |
| AA-Omniscience: non-hallucinationnot in index | 5.5%#517 | 11.0%#396 | 3.9%#535 | Artificial Analysis |
Instruction following
| Benchmark | no reasoning | high | max | Source | Trend |
|---|---|---|---|---|---|
| IFBench | 47.2%#189 | 73.5%#40 | 79.2%#11 | Artificial Analysis |
Long context
| Benchmark | no reasoning | high | max | Source | Trend |
|---|---|---|---|---|---|
| AA-LCR | 41.7%#351 | 72.0%#190 | 74.3%#161 | Artificial Analysis |
Composite
| Benchmark | no reasoning | high | max | Source | Trend |
|---|---|---|---|---|---|
| AA Intelligence Index v4.3.2 | 18.9#245 | 24.4#173 | 24.2#177 | Artificial Analysis |
What a task costs
Estimates from list price, output speed and time to first answer for the best configuration. “With caching” assumes three-quarters of the input is served from the prompt cache. Reasoning tokens are not modelled.
| Workload | Tokens in / out | Cost | With caching | Time |
|---|---|---|---|---|
| Chat reply | 400 / 300 | $0.0001 | – | 2.1 min |
| Summarize a 30-page report | 12,000 / 600 | $0.0015 | – | 2.2 min |
| Code edit | 6,000 / 1,500 | $0.0010 | – | 2.5 min |
| Agentic coding session | 60,000 / 4,000 | $0.0075 | – | 3.4 min |
| Structured extraction | 2,000 / 200 | $0.0003 | – | 2.1 min |
More from DeepSeek
See also
Data as of 22 Oct 2026. Compare these configurations.
Cite as: BenchLeader, “DeepSeek V4 Flash 0420: benchmarks, pricing, speed and rank”, https://www.benchleader.com/models/deepseek-v4-flash-0420, data as of 22 Oct 2026.