VerdictPal
Join

Models / DeepSeek

DeepSeek V4 Flash (Reasoning, High Effort)

DeepSeek V4 · Released 2026-04-24

AA Intelligence Index 37.5. Listed on Artificial Analysis with API pricing $0.14/$0.28 per 1M tokens. Benchmark rows sync from the AA snapshot — see the ledger on this page.

Frontier model tracked on Artificial Analysis. Use the benchmark ledger below for sourced scores; we do not infer numbers beyond published rows.

#87 of 184 on AA Index · current snapshot
25 AA Index · 10 benchmark rows · 1 sourcesPosition is benchmark-specific — not a cross-family or cross-source ranking.
Composite51
Context128K
Input / 1M$0.14
Output / 1M$0.28
Knowledge cutoffopen
Statussolid

Family profile

Best published score in each covered benchmark family.

64/100 avg
Reasoning
87
Coding
40
Agentic
57
Long context
72

4 tested benchmark families

Benchmark placements

Where DeepSeek V4 Flash (Reasoning, High Effort) places on each public benchmark source that publishes a row. Rank counts every model with a latest row in the same test — not a universal quality score.

BenchmarkFamilyRankScoreSourceDate
IFBenchReasoning#35/ 12773Artificial Analysis2026-09-10
τ³-Bench BankingAgentic#47/ 11126Artificial Analysis2026-09-10
Humanity's Last ExamReasoning#77/ 18530Artificial Analysis2026-09-10
SciCodeCoding#77/ 9440Artificial Analysis2026-09-10
Terminal-BenchAgentic#79/ 18957Artificial Analysis2026-09-10
GPQA DiamondReasoning#81/ 19087Artificial Analysis2026-09-10
Artificial Analysis Intelligence IndexReasoning#89/ 18825Artificial Analysis2026-09-10
AA-LCRLong context#105/ 18272Artificial Analysis2026-09-10
AA output speedPerformance#105/ 190Artificial Analysis2026-09-10
AA time to first tokenPerformance#105/ 190Artificial Analysis2026-09-10
Family context

The three highest-scoring models with pages in each capability family. Where DeepSeek V4 Flash (Reasoning, High Effort) shows up, it's highlighted.

ReasoningDeepSeek V4 Flash (Reasoning, High Effort) · 87
  1. 1
    GPT-6 Astra
    96
  2. 2
    Claude Mythos Preview
    95
  3. 3
    Gemini 3.7 Flash
    95
CodingDeepSeek V4 Flash (Reasoning, High Effort) · 40
  1. 1
    GPT 5.5
    96
  2. 2
    Claude Opus 4.8
    96
  3. 3
    Claude Opus 4.6
    96
AgenticDeepSeek V4 Flash (Reasoning, High Effort) · 57
  1. 1
    GPT 5.5
    100
  2. 2
    Claude Opus 4.8
    100
  3. 3
    Claude Fable 5
    100
Long contextDeepSeek V4 Flash (Reasoning, High Effort) · 72
  1. 1
    Kimi K3
    89
  2. 2
    Claude Fable 5.1
    85
  3. 3
    GPT-5.5 (high)
    84
Newest receipts

Benchmark rows added to the public ledger for DeepSeek V4 Flash (Reasoning, High Effort) in the last 120 days. Older rows live in the full table below.

10 recent
BenchmarkFamilyScoreSourceDays ago
AA-LCRLong context72%Artificial Analysis4
AA output speedPerformance0 t/sArtificial Analysis4
AA time to first tokenPerformance0.00sArtificial Analysis4
Artificial Analysis Intelligence IndexReasoning24.8%Artificial Analysis4
GPQA DiamondReasoning86.7%Artificial Analysis4
Humanity's Last ExamReasoning30.3%Artificial Analysis4
IFBenchReasoning73.47%Artificial Analysis4
SciCodeCoding40.2%Artificial Analysis4
τ³-Bench BankingAgentic26.19%Artificial Analysis4
Terminal-BenchAgentic56.93%Artificial Analysis4
Family coverage

A high composite that hides a weak family is a trap. These bars surface the families where this model hasn't been publicly tested, and where it leads.

Reasoning87
Coding40
Agentic57
Long context72
Price vs. performance
0255075100$0.10$0.30$1$3$10$30input USD / 1M tokens · log scale →← intelligence

Hover any dot for name, score, and input price. Keyboard: tab through the top twelve, or use the ranking below.

AnthropicOpenAIMistralAlibabaDeepSeekGoogleMoonshotMetaxAIMiniMaxZhipuXiaomiOtherMeituanCohereNVIDIA

Full benchmark ledger

Every catalog benchmark for DeepSeek V4 Flash (Reasoning, High Effort). Scores link to the original source; gaps mean no public row exists yet.

10 sourced rows

10 of 37 catalog benchmarks have a sourced row for DeepSeek V4 Flash (Reasoning, High Effort).

Reasoning

Coding

Agentic

Long context

Performance

Sources