VerdictPal
Join

Models / DeepSeek

DeepSeek V4 Pro (Reasoning, High Effort)

DeepSeek V4 · Released 2026-04-24

AA Intelligence Index 43.1. Listed on Artificial Analysis with API pricing $0.435/$0.87 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.

#42 of 168 on AA Index · current snapshot
44 AA Index · 10 benchmark rows · 1 sourcesPosition is benchmark-specific — not a cross-family or cross-source ranking.
Composite56
Context128K
Input / 1M$0.43
Output / 1M$0.87
Knowledge cutoffopen
Output speed50 t/s
TTFT (default API)1.16s
Statussolid

Family profile

Best published score in each covered benchmark family.

67/100 avg
Reasoning
91
Coding
46
Agentic
65
Long context
67

4 tested benchmark families

Benchmark placements

Where DeepSeek V4 Pro (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
GPQA DiamondReasoning#38/ 18591Artificial Analysis2026-09-01
τ³-Bench BankingAgentic#43/ 10626Artificial Analysis2026-09-01
IFBenchReasoning#47/ 12771Artificial Analysis2026-09-01
Humanity's Last ExamReasoning#52/ 18035Artificial Analysis2026-09-01
Artificial Analysis Intelligence IndexReasoning#55/ 18744Artificial Analysis2026-09-01
Terminal-BenchAgentic#58/ 18465Artificial Analysis2026-09-01
AA time to first tokenPerformance#59/ 186Artificial Analysis2026-09-01
AA output speedPerformance#66/ 187Artificial Analysis2026-09-01
SciCodeCoding#70/ 18046Artificial Analysis2026-09-01
AA-LCRLong context#111/ 17967Artificial Analysis2026-09-01
Family context

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

ReasoningDeepSeek V4 Pro (Reasoning, High Effort) · 91
  1. 1
    Claude Mythos Preview
    95
  2. 2
    Gemini 3.7 Flash
    95
  3. 3
    Grok 4.6
    95
CodingDeepSeek V4 Pro (Reasoning, High Effort) · 46
  1. 1
    GPT 5.5
    96
  2. 2
    Claude Opus 4.8
    96
  3. 3
    Claude Opus 4.6
    96
AgenticDeepSeek V4 Pro (Reasoning, High Effort) · 65
  1. 1
    GPT 5.5
    100
  2. 2
    Claude Opus 4.8
    100
  3. 3
    Claude Fable 5
    100
Long contextDeepSeek V4 Pro (Reasoning, High Effort) · 67
  1. 1
    Muse Spark 1.2
    83
  2. 2
    Kimi K3
    83
  3. 3
    Gemini 3.5 Flash
    81
Newest receipts

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

10 recent
BenchmarkFamilyScoreSourceDays ago
AA-LCRLong context67%Artificial Analysis6
AA output speedPerformance50 t/sArtificial Analysis6
AA time to first tokenPerformance1.16sArtificial Analysis6
Artificial Analysis Intelligence IndexReasoning43.7%Artificial Analysis6
GPQA DiamondReasoning90.5%Artificial Analysis6
Humanity's Last ExamReasoning35.2%Artificial Analysis6
IFBenchReasoning71.29%Artificial Analysis6
SciCodeCoding46.4%Artificial Analysis6
τ³-Bench BankingAgentic26.19%Artificial Analysis6
Terminal-BenchAgentic64.79%Artificial Analysis6
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.

Reasoning91
Coding46
Agentic65
Long context67
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.

AnthropicOpenAIMistralAlibabaDeepSeekGoogleMoonshotMetaMiniMaxxAIZhipuXiaomiOtherMeituanCohereNVIDIA

Full benchmark ledger

Every catalog benchmark for DeepSeek V4 Pro (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 Pro (Reasoning, High Effort).

Reasoning

Coding

Agentic

Long context

Performance

Sources