Compare / DeepSeek-V4-Pro (0423) vs DeepSeek-V4-Pro (0813)

DeepSeek-V4-Pro (0423) vs DeepSeek-V4-Pro (0813)

Which is cheaper, which is more capable, and which does better at each kind of work, from 17 results on 3 sources that measured both models.

DeepSeekDeepSeek-V4-Pro (0423) Released 24 Apr 2026 · $1.42 in · $2.83 out
DeepSeekDeepSeek-V4-Pro (0813) Released 13 Aug 2026 · $1.32 in · $3.96 out
DeepSeek-V4-Pro (0423) wins 17 too close to callDeepSeek-V4-Pro (0813) wins 9

Of the 17 results both models have, the better value on each. Results whose reported ranges overlap are too close to call.

Results as of 19 November 2026 (catalogue release 2026-11-19-3608133e8331).

DeepSeek-V4-Pro (0423) or DeepSeek-V4-Pro (0813)? The short answer

DeepSeek-V4-Pro (0813) is the newer version at about the same price, and it scores higher (Intelligence Index 36.0 against 30.4): upgrade.

  • They cost about the same: $1.77 and $1.98 per million tokens (three input to one output).
  • DeepSeek-V4-Pro (0813) scores higher on the Artificial Analysis Intelligence Index: 36.0 against 30.4.
  • DeepSeek-V4-Pro (0813) is newer: released 13 Aug 2026, against 24 Apr 2026 for DeepSeek-V4-Pro (0423).
  • On third-party benchmarks, on the results both models have: DeepSeek-V4-Pro (0813) does better for agents and tool use, reasoning and knowledge and sticking to the facts.

Head to head

30.4
Intelligence Index
36.0
$1.77
Price per million tokens, blendedlower is better
$1.98
1,048,576
Context window, tokens
1,048,576

Every fact side by side

DeepSeek-V4-Pro (0423)DeepSeek-V4-Pro (0813)
Price per million tokens$1.42 in · $2.83 out$1.32 in · $3.96 out
Blended (3 in : 1 out)$1.77$1.98
Cheapest host, blended$1.05 (Relace, FP4)$0.87 (Alibaba)
Cached input, per million–$0.044 cached input
Intelligence Index30.436.0
Spring Prompt overall27–
Output speed, tokens a second–96
Context window1,048,576 tokens1,048,576 tokens
Released24 Apr 202613 Aug 2026 (newer)
WeightsOpenOpen
Developer based inChinaChina

Prices are the developer's list price where we have read it, otherwise the typical host on OpenRouter; see every model's price. Intelligence and speed from Artificial Analysis.

Which is better for what

Each model's rank among the models on sale today, for each kind of work, from our own benchmarks and licensed sources. Further right is better; the better rank is in green.

Decks from an analysisTurning a finished analysis into a deck you could present as it is 35 of 35 not measured
Marketing planningPlanning a year of ad spend without overspending 29 of 35 not measured
Agents and tool useMulti-step tasks with tools: support desks, coding agents, function calls 92 of 159 39 of 159
Professional workReal tasks from banking, consulting and law, business documents and freelance projects not measured 29 of 51
Reasoning and knowledgeHard questions across science, maths and general knowledge 51 of 187 35 of 187
WritingWhat people prefer in blind comparisons, and judged writing quality 30 of 169 38 of 169
Sticking to the factsSummarising without inventing things, and factual answers 56 of 139 44 of 139
Speed under pressureGood decisions against a real clock (fast chess) 39 of 39 not measured

Every shared result

Each model's best result on every measure both have, grouped by benchmark, with the setting that produced it. Under each measure, every model that benchmark has measured is a grey tick, best to the right, so you can see where the two sit in the field. The headline measures of each benchmark come first; the rest are under "Show all". Green marks the better value; where a source reports ranges that overlap, neither is marked.

MeasureDeepSeek-V4-Pro (0423)DeepSeek-V4-Pro (0813)
Arena (formerly LMArena) reported by Arena (formerly LMArena) · 3 results
Overall · TextArena rating, higher is better 53 · 48 of 387 measured 1,458 1,464high reasoning
Overall · Text factualityArena rating, higher is better 49 · 50 of 163 measured 1,455 1,454high reasoning
Business, management and finance · TextArena rating, higher is better 52 · 37 of 380 measured 1,460 1,467high reasoning
Artificial Analysis reported by Artificial Analysis · 3 results
Artificial Analysis Intelligence Indexindex score, higher is better 53 · 42 of 249 measured 30.4max reasoning 36.0max reasoning
GPQA Diamond% of questions, higher is better 37 · 15 of 232 measured 90.5%high reasoning 92.8%max reasoning
Humanity's Last Exam% of questions, higher is better 57 · 42 of 250 measured 37.5%max reasoning 41.0%max reasoning
SimpleQA Verified (Epoch AI) reported by SimpleQA Verified (Epoch AI) · 1 result
Correct answers% of questions, higher is better 32 · 20 of 76 measured 47.0%max reasoning 52.9%max reasoning
Show all 17 results (10 more)
MeasureDeepSeek-V4-Pro (0423)DeepSeek-V4-Pro (0813)
Arena (formerly LMArena) reported by Arena (formerly LMArena) · 4 results
Creative writing · TextArena rating, higher is better 39 · 44 of 385 measured 1,446 1,445high reasoning
Expert prompts · TextArena rating, higher is better 64 · 57 of 338 measured 1,480 1,484high reasoning
Instruction following · TextArena rating, higher is better 52 · 40 of 387 measured 1,452 1,461high reasoning
Writing, literature and language · TextArena rating, higher is better 44 · 39 of 386 measured 1,451 1,454high reasoning
Artificial Analysis reported by Artificial Analysis · 6 results
Terminal-Bench 2.1% of tasks, higher is better 42 · 26 of 131 measured 64.8%high reasoning 78.7%max reasoning
Terminal-Bench 4.0% of tasks, higher is better 32 · 34 of 120 measured 14.6%max reasoning 14.1%max reasoning
τ-bench banking% of tasks, higher is better 35 · 19 of 121 measured 30.1%max reasoning 39.6%max reasoning
Artificial Analysis Coding Indexindex score, higher is better 40 · 30 of 130 measured 59.4max reasoning 68.8max reasoning
Long-context reasoning (AA-LCR)% of questions, higher is better 86 · 39 of 241 measured 74.7%max reasoning 80.3%max reasoning
SciCode% of problems, higher is better 50 · 49 of 121 measured 50.8%max reasoning 51.0%max reasoning

Quick answers

Is DeepSeek-V4-Pro (0423) better than DeepSeek-V4-Pro (0813)?

DeepSeek-V4-Pro (0813) is the newer version at about the same price, and it scores higher (Intelligence Index 36.0 against 30.4): upgrade.

Which is cheaper, DeepSeek-V4-Pro (0423) or DeepSeek-V4-Pro (0813)?

They cost about the same: $1.77 and $1.98 per million tokens (three input to one output).

Which is better for agents and tool use, DeepSeek-V4-Pro (0423) or DeepSeek-V4-Pro (0813)?

DeepSeek-V4-Pro (0813) ranks 39 of 159 models on sale for multi-step tasks with tools: support desks, coding agents, function calls, against 92 for DeepSeek-V4-Pro (0423).

Which is better for reasoning and knowledge, DeepSeek-V4-Pro (0423) or DeepSeek-V4-Pro (0813)?

DeepSeek-V4-Pro (0813) ranks 35 of 187 models on sale for hard questions across science, maths and general knowledge, against 51 for DeepSeek-V4-Pro (0423).

Which is better for writing, DeepSeek-V4-Pro (0423) or DeepSeek-V4-Pro (0813)?

DeepSeek-V4-Pro (0423) ranks 30 of 169 models on sale for what people prefer in blind comparisons, and judged writing quality, against 38 for DeepSeek-V4-Pro (0813).

Which is better for sticking to the facts, DeepSeek-V4-Pro (0423) or DeepSeek-V4-Pro (0813)?

DeepSeek-V4-Pro (0813) ranks 44 of 139 models on sale for summarising without inventing things, and factual answers, against 56 for DeepSeek-V4-Pro (0423).

Run DeepSeek-V4-Pro (0423) and DeepSeek-V4-Pro (0813) on your own prompt.

Benchmarks aren't your task. Test both side by side on your prompt, with every answer checked against your rules and the cost of each one.