DeepSeek Models

33 modelsGeneral models from $0.154/M inputUp to 1M contextOfficial site ↗

Usage

Last 29 days · 2026-08-17 to 2026-09-14

Tokens

561K

Requests

5K

Models in use

6 of 33

Tokens per day, stacked by model

027.3K54.7K08-1708-2408-3109-0709-142026-08-17 — 15,740 tokens deepseek-v3.2: 15,100 deepseek-v4-pro: 6402026-08-18 — 17,000 tokens deepseek-v3.2: 16,435 deepseek-v4-pro: 5652026-08-19 — 18,125 tokens deepseek-v3.2: 18,1252026-08-20 — 10,705 tokens deepseek-v3.2: 10,7052026-08-21 — 15,940 tokens deepseek-v3.2: 15,9402026-08-22 — 18,080 tokens deepseek-v3.2: 18,0802026-08-23 — 15,200 tokens deepseek-v3.2: 15,2002026-08-24 — 14,255 tokens deepseek-v3.2: 14,2552026-08-25 — 13,440 tokens deepseek-v3.2: 13,4402026-08-26 — 15,760 tokens deepseek-v3.2: 15,7602026-08-27 — 14,050 tokens deepseek-v3.2: 14,0502026-08-28 — 14,000 tokens deepseek-v3.2: 14,0002026-08-29 — 12,840 tokens deepseek-v3.2: 12,8402026-08-30 — 14,230 tokens deepseek-v3.2: 14,2302026-08-31 — 15,925 tokens deepseek-v3.2: 15,9252026-09-01 — 15,275 tokens deepseek-v3.2: 15,2752026-09-02 — 15,755 tokens deepseek-v3.2: 15,7552026-09-03 — 15,475 tokens deepseek-v3.2: 15,4752026-09-04 — 16,495 tokens deepseek-v3.2: 16,4952026-09-05 — 18,365 tokens deepseek-v3.2: 18,3652026-09-06 — 16,360 tokens deepseek-v3.2: 16,3602026-09-07 — 16,360 tokens deepseek-v3.2: 16,3602026-09-08 — 49,645 tokens deepseek-v4-pro: 33,115 deepseek-v3.2: 16,5302026-09-09 — 17,820 tokens deepseek-v3.2: 17,8202026-09-10 — 30,310 tokens deepseek-v3.2: 26,435 deep-deepseek-v4-flash: 1,855 deepseek-v4-pro: 1,595 deep-deepseek-v4-pro: 4252026-09-11 — 34,250 tokens deepseek-v3.2: 33,830 deep-deepseek-v4-flash: 275 deepseek-v4-pro: 1452026-09-12 — 17,150 tokens deepseek-v3.2: 17,1502026-09-13 — 17,495 tokens deepseek-v3.2: 17,4952026-09-14 — 54,690 tokens deepseek-v3.2: 54,690
  • deepseek-v3.2
  • deepseek-v4-pro
  • deep-deepseek-v4-flash
  • deep-deepseek-v4-pro

Which models that traffic went to

  1. DeepSeek V3.293.1%522K
  2. DeepSeek V4 Pro6.4%36.1K
  3. Deep Deepseek V4 Flash0.4%2.1K
  4. Deep Deepseek V4 Pro0.1%425

Share of 561K tokens. 2 models with traffic report no token counts and cannot be ranked here, including DeepSeek-V3.2-Exp and deepseek-r1-distill-llama-70b — they are in the request view.

The two views disagree on purpose: a model can take a large share of the calls and a small share of the tokens — many short requests — or the reverse. Which one matters depends on whether your cost is driven by call volume or by prompt length. Measured on AIHubMix over the last 29 days, counting the 33 model IDs listed on this page; traffic routed through upstream-specific IDs that are not in the public catalog is not included.

All 33 DeepSeek Models

Open in model list
DeepSeek models on AIHubMix with input and output modalities, context length, maximum output, price per million tokens including cache read and cache write rates, and measured throughput and latency.
Modalities
deepseek-v4-flashTakes text, returns text.1M$0.464$0.928/M$0.0093/M59 tok/s0.69 s
deepseek-v4-proTakes text, returns text.1M$0.464$0.928/M$0.0039/M59 tok/s0.69 s
deepseek-v3.2Takes text, returns text.164K$0.274$0.411/M$0.0274/M59 tok/s0.69 s
DeepSeek-V3.1-ThinkTakes text, returns text.164K$0.56$1.68/M—32 tok/s1.29 s
DeepSeek-V3.1-FastTakes text, returns text.164K$1.096$3.288/M—150 tok/s0.20 s
DeepSeek-V3.2-ExpTakes text, returns text.131K$0.274$0.411/M$0.0274/M45 tok/s0.20 s
DeepSeek-V3.2-Exp-ThinkTakes text, returns text.131K$0.274$0.411/M$0.0274/M23 tok/s1.96 s
deepseek-r1-distill-llama-70bTakes text, returns text.131K$0.8$1.6/M———
DeepSeek-V3Takes text, returns text.128K$0.272$1.088/M—67 tok/s0.59 s
deepseek-v3.1-terminusTakes text, returns text.128K$0.571$1.714/M—50 tok/s0.20 s
DeepSeek-OCRTakes text, vision. Output modality not published.8K$0.0411$0.1644/M———
deepseek-ai/DeepSeek-R1-Distill-Llama-8B—$0.01$0.01/M———
deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B—$0.01$0.01/M———
deepseek-ai/DeepSeek-R1-Distill-Qwen-7B—$0.01$0.01/M———
tngtech/DeepSeek-R1T-Chimera—$0.02$0.02/M———
DeepSeek-R1-Distill-Qwen-7B—$0.06$0.12/M———
deepseek-ai/DeepSeek-Prover-V2-671B—$0.1$0.1/M———
deepseek-ai/DeepSeek-R1-Distill-Qwen-14B—$0.1$0.1/M———
deep-deepseek-v4-flashTakes text, returns text.—$0.154$0.308/M$0.0013/M59 tok/s0.69 s
deepseek-ai/DeepSeek-Coder-V2-Instruct—$0.16$0.32/M———
deepseek-ai/deepseek-llm-67b-chat—$0.16$0.16/M———
deepseek-ai/DeepSeek-V2-Chat—$0.16$0.32/M———
deepseek-ai/DeepSeek-V2.5—$0.16$0.32/M———
deepseek-ai/deepseek-vl2—$0.16$0.16/M———
deepseek-ai/DeepSeek-R1-Distill-Qwen-32B—$0.2$0.2/M———
baidu-deepseek-v3.2Takes text, returns text.—$0.274$0.411/M———
drun-deepseek-v3.2Takes text, returns text.—$0.274$0.411/M$0.0274/M——
DeepSeek-R1-Distill-Qwen-32B—$0.28$0.84/M———
deep-deepseek-v4-proTakes text, returns text.—$0.464$0.928/M$0.0039/M59 tok/s0.69 s
DeepSeek-V3-FastTakes text, returns text.—$0.56$2.24/M—150 tok/s1.46 s
deepseek-ai/DeepSeek-R1-Distill-Llama-70B—$0.6$0.6/M———
deepseek-ai/Janus-Pro-7B—$2$2/M———
deepseek-ai/DeepSeek-R1-Zero—$2.2$2.2/M———

Prices are USD per million tokens; cache read and cache write are the rates for prompt-cache hits and for writing a prompt into the cache. Throughput and latency are measured on AIHubMix — the same figures the model detail page shows — not vendor claims. A dash means the catalog does not publish that field for that model, which is not the same as the model not supporting it.

Common Comparisons

baidu-deepseek-v3.2 vs deep-deepseek-v4-flashbaidu-deepseek-v3.2 vs deep-deepseek-v4-probaidu-deepseek-v3.2 vs deepseek-r1-distill-llama-70bbaidu-deepseek-v3.2 vs gpt-5.6-lunaahm-Phi-3-5-vision-instruct vs baidu-deepseek-v3.2

DeepSeek on AIHubMix

Which DeepSeek model should I start with?

deep-deepseek-v4-flash at $0.154/M input — the cheapest entry here that declares tool calling. Move up to deepseek-ai/DeepSeek-R1-Zero when answer quality matters more than cost, or to DeepSeek-V3.1-Think for long-form reasoning.

Which of these models reason before answering?

3 of the 33 models here declare a reasoning phase — they work through the problem before producing an answer, which helps on multi-step problems at the cost of extra output tokens. Use the Reasoning filter above the table to see them. The catalog does not record anything further about how they differ, so this page does not sort them into families.

Why are there several entries for the same model?

Because each row is a route you can call, not a model release. Some IDs name an upstream (azure-, alicloud-, cc-), some are the open-weight repository form (deepseek-ai/…), and some differ only in capitalisation, kept so older integrations keep working.

The catalog does not carry a field saying which of those a given row is, so this page does not sort them into buckets it would have to invent. Every row shows that route’s own price, context and speed — compare those directly, and open a model to see the upstreams that serve it.

How is cached input billed?

The Cache read column is the rate for input tokens served from the prompt cache — for example deep-deepseek-v4-flash bills cache hits at 0.83% of the input rate and deep-deepseek-v4-pro bills cache hits at 0.83% of the input rate. Cache write is the surcharge for putting a prompt into the cache in the first place, and only a few upstreams bill it separately. A dash in either column means the catalog carries no cache rate for that model, so plan on paying the full input rate.

Do I need a separate DeepSeek account?

No. One AIHubMix key covers every model on this page, and switching between them is a change to the model string — billing, rate limits, and logs stay in one place.

Other Model Authors

See all model authors →
OpenAI
145 modelsQwen
118 modelsGoogle
97 modelsLlama
42 modelsChatGLM
40 models

Start calling DeepSeek in one line

One key, one endpoint, 750 models across 29 model authors.

Get API keyRead the docs

© 2023 - 2026 AIHubMix, LLC

Product

Developer

Connect

Terms & Policies