Skip to content
VECTOR WIREAI INTELLIGENCE
⌘K
UTC
Models739Deals30d257Policy30d64, 14 pendingSources1,010

DeepSeek-V2.5

DeepSeekReleased Sep 6, 2024Unpriced

Score basis

DeepSeek-V2.5 is behind the leaders in coding. Too few results yet to rate reasoning, agentic tasks, safety, long context, math, multimodal tasks, multilingual tasks, instruction following, or factuality.

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.

Limited1 capability
  1. Coding−45.1%1 of 10
Not rated9 capabilities

Too few results yet to rate Reasoning, Agentic, Safety, Long Context, Math, Multimodal, Multilingual, Instruction Following or Factuality.

Price

No current price is tracked for this model. See the rate card

Evidence

45results on32benchmarks

  • 2 independently verified
  • 3 aggregator
  • 31 vendor-reported
  • 9 cross-referenced

From 6 sources · latest Oct 7, 2026 · How verification works

DeepSeek-V2.5 benchmark results

45 results on 32 benchmarks, grouped by capability. Each bar is the result as a share of the capability leader's; every result links to its source.

Coding

LimitedFull coding ranking

45.1% behind the leader1 of 10 ranked benchmarks measured

Show 1 more coding resultHide 1 coding result

Reasoning

Not enough dataFull reasoning ranking

0 of 6 ranked benchmarks measured

Long Context

Not enough dataFull long context ranking

0 of 3 ranked benchmarks measured

More results

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

Show 34 more resultsHide 34 results

DeepSeek-V2.5: common questions

Who makes DeepSeek-V2.5?

DeepSeek-V2.5 is made by DeepSeek.

When was DeepSeek-V2.5 released?

DeepSeek-V2.5 was released on Sep 6, 2024, according to Artificial Analysis.

What is DeepSeek-V2.5 good at?

DeepSeek-V2.5 is behind the leaders in coding. Too few results yet to rate reasoning, agentic tasks, safety, long context, math, multimodal tasks, multilingual tasks, instruction following, or factuality.

How many benchmarks has DeepSeek-V2.5 been tested on?

We track 45 results for DeepSeek-V2.5 on 32 benchmarks from 6 sources, 2 of them independently verified. The latest was recorded on Oct 7, 2026.

About this record

Where DeepSeek-V2.5's numbers come from, and every name it appears under.

Tracked since
May 2, 2026
Newest source mention
Sep 2, 2026

Where the results come from

Verification: 45 scores · 2 independently verified · 3 aggregator-attributed · 9 vendor cross-reference · 31 vendor-reported. How these tiers are assigned

From 6 sources on 5 sites. arxiv.org supplies 21 of them; the 2 independently verified results come from 2 sites. Bars are coloured by trust tier.

  • arxiv.org21
  • api.llm-stats.com10
  • huggingface.co9
  • artificialanalysis.ai4
  • aider.chat1

Also known as

deepseek chat v2.5deepseek-v2.5 (dec '24)DeepSeek-V2.5 (Dec)DeepSeek-V2.5 (Non-Reasoning)DeepSeek-V2.5 (Reasoning)DeepSeek-V2.5-0905deepseekv2.5 chat