DeepSeek
Explore all 13 AI models from DeepSeek. Compare benchmarks, pricing, and capabilities across the full model lineup.
deepseek.com13
Total Models
79.3
Avg Benchmark
13
Open Source
3
Modalities
All Models
13 models
DeepSeek V4
DeepSeek
DeepSeek's 1T parameter coding-focused model with 1M+ context. Three architectural innovations: Manifold-Constrained Hyper-Connections, Engram memory, Sparse Attention.
Input
$0.10/M
Output
$0.40/M
Context
1.0M
DeepSeek-V3.2
DeepSeek
Unified reasoning and non-reasoning model that merges DeepSeek-V3 and R1 capabilities into a single architecture.
Input
$0.28/M
Output
$0.42/M
Context
128K
DeepSeek-Math V2
DeepSeek
Math-specialized model achieving gold-level scores in math competitions. Based on V3.2 architecture.
Input
$0.27/M
Output
$1.10/M
Context
128K
DeepSeek-V3.1
DeepSeek
Hybrid model combining V3 and R1 strengths. Improved reasoning with RL techniques from R1.
Input
$0.27/M
Output
$1.10/M
Context
128K
DeepSeek-R1-Distill-Llama-70B
DeepSeek
R1's reasoning capability distilled into a Llama 3.1 70B architecture for efficient deployment.
Input
$0.18/M
Output
$0.18/M
Context
128K
DeepSeek-R1
DeepSeek
DeepSeek's reasoning model with transparent chain-of-thought. Open-source and highly competitive.
Input
$0.55/M
Output
$2.19/M
Context
128K
DeepSeek-R1-Distill-Qwen-32B
DeepSeek
R1 reasoning capabilities distilled into a compact Qwen-based 32B model.
Input
$0.12/M
Output
$0.18/M
Context
128K
DeepSeek-R1-Distill-Qwen-7B
DeepSeek
Distilled R1 reasoning into compact Qwen-based model. Exceptional at math and programming.
Input
$0.07/M
Output
$0.14/M
Context
128K
DeepSeek-R1-Distill-Llama-8B
DeepSeek
R1 reasoning distilled into Llama 3 architecture. Strong reasoning at minimal compute cost.
Input
$0.07/M
Output
$0.14/M
Context
128K
DeepSeek-V3
DeepSeek
DeepSeek's open-source MoE model rivaling frontier models at a fraction of the cost.
Input
$0.27/M
Output
$1.10/M
Context
128K
DeepSeek-VL2
DeepSeek
Vision-language model for image understanding, OCR, and visual reasoning tasks.
Input
$0.14/M
Output
$0.28/M
Context
128K
DeepSeek-V2.5
DeepSeek
Merged general and coder capabilities from V2 into a unified model.
Input
$0.14/M
Output
$0.28/M
Context
128K
DeepSeek-Coder-V2
DeepSeek
DeepSeek's open-source code-focused MoE model. Competitive with GPT-4 Turbo on coding.
Input
$0.14/M
Output
$0.28/M
Context
128K