跳转到主要内容

DeepSeek R1 Zero

DeepSeekDeepSeek开源权重MIT · 商用许可

描述

DeepSeek-R1-Zero, a model trained via large-scale reinforcement learning (RL) without supervised fine-tuning (SFT) as a preliminary step, demonstrated remarkable performance on reasoning. With RL, DeepSeek-R1-Zero naturally emerged with numerous powerful and interesting reasoning behaviors. However, DeepSeek-R1-Zero encounters challenges such as endless repetition, poor readability, and language mixing. To address these issues and further enhance reasoning performance, we introduce DeepSeek-R1, which incorporates cold-start data before RL. DeepSeek-R1 achieves performance comparable to OpenAI-o1 across math, code, and reasoning tasks.

发布日期
2025-01-20
参数规模
671.0B
上下文长度
164K
支持模态
text

能力雷达图

80
general
50
coding
90
reasoning
60
science估算
78
agents
0
multimodal

缺少专门科学评测时,Science 由 LLM Stats 科学得分或推理能力估算。

排行榜排名

暂无排名数据

基准测试分数 (LLM Stats)

(LLM Stats (zeroeval))

Math

MATH-50095.9%自报
AIME 202486.7%自报

Reasoning

GPQANYU + Cohere + Anthropic (2023)73.3%自报
LiveCodeBench50.0%自报

AA 评测指数

(Artificial Analysis)

暂无 AA 评测数据

LLM Stats 分类评分

(LLM Stats (zeroeval))
Math
90
Reasoning
80
General
80
Physics
70
Biology
70
Chemistry
70
Code
50

定价

输入价格$0.2574 / 1M tokens
输出价格$1.0287 / 1M tokens
混合价格(3:1)$0.45023 / 1M tokens

速度

暂无速度数据

供应商价格排行

供应商价格排行

1 个供应商

供应商输入输出
1DeepSeek主要
$0.2574
$1.0287

比较该模型在不同 API 供应商之间的定价。

外部链接