Skip to content
VECTOR WIREAI INTELLIGENCE
⌘K
UTC
Models762Deals30d246Policy30d61, 12 pendingSources1,061

DeepSeek-V4-Pro 0813

DeepSeekWeights published Aug 13, 2026Current

Score basis

DeepSeek-V4-Pro 0813 is capable in long context, reasoning, and agentic tasks; and behind the leaders in factuality, coding, math, and instruction following. Too few results yet to rate safety, multimodal tasks, or multilingual tasks.

Capability profile

Bars show the model's median result as a share of the leading model's, per capability. Select a row to see its results.

Capable3 capabilities
  1. Long Context−10.5%1 of 3
  2. Reasoning−15.4%5 of 6
  3. Agentic−22.6%3 of 7
Limited4 capabilities
  1. Factuality−28.7%3 of 4
  2. Coding−31.2%4 of 10
  3. Math−36.0%3 of 5
  4. Instruction Following−41.6%1 of 3
Not rated3 capabilities

Too few results yet to rate Safety, Multimodal or Multilingual.

Price

$1.32input$3.96outputper million tokens

From Artificial Analysis · 6 providers tracked · All prices

Costs more than 72% of 329 priced models · 3:1 input-to-output blend, log scale

Evidence

62results on40benchmarks

  • 18 independently verified
  • 25 aggregator
  • 11 vendor-reported
  • 8 cross-referenced

From 10 sources · latest Oct 17, 2026 · How verification works

API features

Tool callingStructured outputsReasoning

As listed by OpenRouter

DeepSeek-V4-Pro 0813 benchmark results

62 results on 40 benchmarks, grouped by capability. Each bar is the result as a share of the capability leader's; every result links to its source.

Long Context

CapableFull long context ranking

10.5% behind the leader1 of 3 ranked benchmarks measured

Show 1 more long context resultHide 1 long context result

Reasoning

CapableFull reasoning ranking

15.4% behind the leader5 of 6 ranked benchmarks measured

Show 5 more reasoning resultsHide 5 reasoning results

Agentic

CapableFull agentic ranking

22.6% behind the leader3 of 7 ranked benchmarks measured

Show 2 more agentic resultsHide 2 agentic results

Factuality

LimitedFull factuality ranking

28.7% behind the leader3 of 4 ranked benchmarks measured

Show 2 more factuality resultsHide 2 factuality results

Coding

LimitedFull coding ranking

31.2% behind the leader4 of 10 ranked benchmarks measured

Show 3 more coding resultsHide 3 coding results

Math

LimitedFull math ranking

36.0% behind the leader3 of 5 ranked benchmarks measured

Instruction Following

LimitedFull instruction following ranking

41.6% behind the leader1 of 3 ranked benchmarks measured

More results

Benchmarks outside the capability baskets. They are not ranked against a leader.

Show 23 more resultsHide 23 results

DeepSeek-V4-Pro 0813: common questions

Who makes DeepSeek-V4-Pro 0813?

DeepSeek-V4-Pro 0813 is made by DeepSeek.

When was DeepSeek-V4-Pro 0813 released?

DeepSeek-V4-Pro 0813's weights were first published on Hugging Face on Aug 13, 2026.

What is DeepSeek-V4-Pro 0813 good at?

DeepSeek-V4-Pro 0813 is capable in long context, reasoning, and agentic tasks; and behind the leaders in factuality, coding, math, and instruction following. Too few results yet to rate safety, multimodal tasks, or multilingual tasks.

How much does DeepSeek-V4-Pro 0813 cost?

DeepSeek-V4-Pro 0813 costs $1.32 per million input tokens and $3.96 per million output tokens, according to Artificial Analysis. We track its price at 6 providers. At a mix of three input tokens to one output token, it costs more than 72% of the 329 priced models we track.

How many benchmarks has DeepSeek-V4-Pro 0813 been tested on?

We track 62 results for DeepSeek-V4-Pro 0813 on 40 benchmarks from 10 sources, 18 of them independently verified. The latest was recorded on Oct 17, 2026.

Which API features does DeepSeek-V4-Pro 0813 support?

OpenRouter lists tool calling, structured outputs, and reasoning for DeepSeek-V4-Pro 0813.

About this record

Where DeepSeek-V4-Pro 0813's numbers come from, and every name it appears under.

Tracked since
Aug 12, 2026
Newest source mention
Sep 30, 2026

Where the results come from

Verification: 62 scores · 18 independently verified · 25 aggregator-attributed · 8 vendor cross-reference · 11 vendor-reported. How these tiers are assigned

From 10 sources on 6 sites. Artificial Analysis supplies 25 of them; the 18 independently verified results come from 3 sites. Bars are coloured by trust tier.

  • artificialanalysis.ai25
  • huggingface.co11
  • api.llm-stats.com8
  • arcprize.org8
  • livebench.ai7
  • epoch.ai3

Also known as

How our sources name DeepSeek-V4-Pro 0813 at each reasoning setting.

SettingShort formLong form
lowdeepseek v4 pro 0813 (low)—
highdeepseek v4 pro 0813 (high)—
maxdeepseek v4 pro 0813 (max)deepseek v4 pro 0813 (reasoning, max effort)
Also listed asdeepseek v4 pro 0813 (none)deepseek-v4-pro-0813:batchdeepseek 4 prounsloth/deepseek-v4-pro-0813-ggufdeepseek v4 pro 0813 (non-reasoning)deepseek-ai/deepseek-v4-pro-0813