Plain-language provider view: what this company is strongest at, how many active models it has, what pricing we can verify, and which current models matter most.
16
Models
#177
Top Rank
60.2
Avg Capability
331.0M
Downloads
Pricing Posture
8 / 16 models have official company pricing
This tells you how often we can verify direct first-party pricing instead of only broker or router pricing.
Lowest Verified Entry
Free/M
The lowest reliable public price we could verify across this provider's active lineup.
Strategic Strength
LLMs
DeepSeek-V3 leads current estimated value at $60M.
Deployment Reach
11 / 16 models have verified deploy or runtime access
9 on your computer · 1 cloud servers you control · 4 hosted for you
Open weights do not always mean easy hosted access. For DeepSeek, they usually mean you bring the hardware yourself or rent a cloud GPU when the models are larger.
Most open models here can run on your own hardware.
11 easy personal-hardware fits · 0 desktop GPU fits · 0 cloud GPU fits · 2 high-memory cloud fits
Pricing
1
Open Source
2
Recent launches, pricing moves, benchmark updates, API changes, and research signals linked to this provider.
Recent updates about new ways to use this provider's models, including self-host and official runtime options.
R1 is now available through local Ollama runtime. 128K context window listed. A version of the DeepSeek-R1 model that has been post trained to provide unbiased, accurate, and factual information by Perplexity.
| # | Model | Category | Params | Capability | Downloads | Cheapest Verified | Est. Value | Open |
|---|---|---|---|---|---|---|---|---|
| 177 | deepseek-v3.1 | LLMs | Unknown |
50.0K
Likes
13
Open Models
R1 is now available through local Ollama runtime. 128K context window listed. A version of the DeepSeek-R1 model that has been post trained to provide unbiased, accurate, and factual information by Perplexity.
View sourceDeepSeek-R1 is now available through local Ollama runtime. 128K context window listed. A version of the DeepSeek-R1 model that has been post trained to provide unbiased, accurate, and factual information by Perplexity.
View sourceDeepSeek-R1 is now available through local Ollama runtime. 128K context window listed. A version of the DeepSeek-R1 model that has been post trained to provide unbiased, accurate, and factual information by Perplexity.
View source| 67.0 |
| 499.0K |
Custom Start Free Trial |
| $43M |
| 72 | DeepSeek-V3 | LLMs | 685B | 76.5 | 320.0M | Custom Start Free Trial | $60M |
| 183 | DeepSeek-R1 | LLMs | 671B | 81.6 | 4.5M | Custom Start Free Trial | $59M |
| 230 | DeepSeek-V4-Pro | LLMs | Unknown | 81.1 | 1.2M | Custom Deploy | $31M |
| 233 | DeepSeek-V4-Flash | LLMs | Unknown | 79.3 | 2.1M | Custom Deploy | $30M |
| 282 | DeepSeek-V3.2 | LLMs | Unknown | 62.8 | 869.0K | Custom Start Free Trial | $42M |
| 13 | DeepSeek V3.2 Speciale | LLMs | Unknown | 79.1 | 0 | Custom Start Free Trial | $42M |
| 36 | DeepSeek V3.1 Terminus | LLMs | Unknown | 63.0 | 0 | Custom Start Free Trial | $39M |
| 235 | R1 Distill Llama 70B | LLMs | 70B | 59.9 | 0 | Custom Start Free Trial | $38M |
| 458 | DeepSeek-V4-Flash-DSpark | LLMs | Unknown | 43.2 | 40.3K | Free Free | $18M |
| 390 | DeepSeek-V4-Pro-DSpark | LLMs | Unknown | 43.2 | 10.3K | Free Free | $18M |
| 86 | R1 Distill Qwen 32B | LLMs | 32B | 62.5 | 0 | Custom Start Free Trial | $40M |
| 316 | DeepSeek-OCR-2 | Specialized | Unknown | 56.8 | 1.7M | Custom Start Free Trial | $25M |
| 590 | DeepSeek-V4-Flash-Base | Specialized | Undisclosed | 36.8 | 8.2K | — | $17M |
| 595 | DeepSeek-V4-Pro-Base | Specialized | Undisclosed | 36.8 | 3.2K | — | $16M |
| 1334 | dspark_gemma4_12b_block7 | Specialized | 12B | 34.1 | 591 | — | --- |