Inference
EN
DeepSeek

DeepSeek-V2.5

DeepSeek-V2.5 is a standard model from DeepSeek. It has 2 scored results in 2 categories. It has no index.

  • Open weights
  • Standard
  • Input: text
  • Released 2024-09
UnrankedEnough results for an index.
IndexAll categories, out of 100.UnrankedMedian 51.0
CoverageIndex weight that has results.35%
SpeedOutput tokens per second.—Median 67/s
First tokenSeconds to the first output token.—Median 1.3 s
Input / 1MUS dollars per 1M input tokens.—Median $0.60
Output / 1MUS dollars per 1M output tokens.—Median $2.55
ContextMaximum tokens in one request.128KMedian 262K
EloLMArena rating and rank.1271±5 #285Median 1422

Capabilities

Index per category, out of 100.
Out of 100
AgenticMulti-step tasks with tools.None counted——
CodingCode writing and repair.1 counted29.5—
ReasoningLogic problems and puzzles.None counted——
MultimodalTasks with images and text.None counted——
KnowledgeFacts and expert knowledge.1 counted16.0—
MultilingualTasks in many languages.None counted——
InstructionTasks with strict prompt rules.None counted——
MathMath problems.None counted——

Results

2 counted
BenchmarkThe benchmark name.ResultThe score from the source.PlacePlace among results on this benchmark.
Evaluating Large Language Models Trained on CodeCoding89.0%6 / 22
Artificial Analysis GPQA DiamondKnowledge42.3%187 / 199

Sources

BenchLMCC BY-NC 4.0

Questions

What is the context window of DeepSeek-V2.5?
128K tokens.
Is DeepSeek-V2.5 open source?
Yes. Its weights are published.
When was DeepSeek-V2.5 released?
September 2024.

More from DeepSeek

All models
DeepSeek V329.9DeepSeek-R138.9DeepSeek R1 Distill Qwen 32BUnrankedR1 Distill Llama 70BUnrankedDeepSeek V3 032437.1