Цена входных данных
15$0.300/M
15$0.300/M
16DeepSeek V3$0.400/M
16$0.400/M
DeepSeek V3 is available through 149 API providers on LMSpeed. Compare API pricing from $0.0007 to $547.50 per million input tokens across providers. Free API access is offered by 2 providers. In speed benchmarks, the fastest provider reaches 345 tok/s.
DeepSeek V3 is DeepSeek flagship MoE language model with 671B total parameters, delivering strong performance in reasoning, coding, and multilingual tasks at competitive inference cost.
Также известна как
Compare DeepSeek V3 API pricing across 147 providers. Input prices range from $0.0007 to $547.50 per million input. MMKG offers the lowest rate at $0.0007/M. 2 providers offer free API credits or a free tier.
| Провайдер | Вариант модели | Аудит | Входные данные ($/M) | Выходные данные ($/M) | Скорость (т/с) | Первый токен |
|---|---|---|---|---|---|---|
AIGCBAR99.1% | deepseek-v3 | — | $0.517 | $2.06 | 84.0 t/s | 8.98 s |
丸美小沐99.8% | deepseek-v3 | — | $75.00 | $75.00 | 26.7 t/s | 3.41 s |
老张API100% | deepseek-v3-0324 | — | $0.240-40% | $0.959 | — | — |
| deepseek-v3-20250324 | — | $0.240-40% | $0.959 | — | — | |
| deepseek-v3-250324 | — | $0.240-40% | $0.959 | — | — | |
| deepseek-v3 | — | $0.274-31% | $1.10 | — | — | |
柏拉图AI100% | deepseek-v3 | — | $0.274-32% | $0.274-69% | — | — |
V-API100% | deepseek-v3 | — | $2.00 | $8.00 | — | — |
天絮 API100% | Pro/deepseek-ai/DeepSeek-V3 | — | $1.79 | $7.14 | — | — |
| deepseek-ai/DeepSeek-V3 | — | $8.93 | $8.93 | — | — | |
MapleLeaf API100% | deepseek-v3 | — | $0.287-28% | $1.15 | — | — |
速创API99.9% | deepseek-v3 | — | $2.00 | $2.00 | — | — |
6345ywz API99.8% | deepseek/deepseek-v3-0324 | — | $0.037-91% | $0.153-83% | — | — |
| deepseek-ai/DeepSeek-V3 | — | $0.171-57% | $0.171-81% | — | — | |
初叶🍂Furry API98.1% | deepseek-v3 | — | $0.050-87% | $0.050-94% | — | — |
| siliconflow/deepseek-v3-0324 | — | $0.050-87% | $0.050-94% | — | — | |
| vanchin/deepseek-v3 | — | $0.050-87% | $0.050-94% | — | — | |
素墨API100% | deepseek-v3-0324 | — | $0.010-97% | $0.010-99% | — | — |
CookingAI100% | deepseek-v3-0324 | — | $0.010-97% | $0.010-99% | — | — |
MMKG100% | deepseek-chat-search | — | $0.0007-100% | $0.0007-100% | — | — |
Данные о ценах из публичных API провайдеров
Сравните производительность по скорости и задержке у всех API-провайдеров.
| Провайдер | Скорость | Задержка | Тесты |
|---|---|---|---|
Kriora deepseek-v3-0324 | 344.79 tok/s | 0.82s | 15 |
deepseek-ai/DeepSeek-V3-0324-fast | 210.41 tok/s | 2.42s | 5 |
DeepSeek-V3-0324 | 108.64 tok/s | 5.11s | 20 |
DeepSeek-V3-Fast | 95.48 tok/s | 0.52s | 30 |
deepseek-v3 | 83.97 tok/s | 8.98s | 5 |
deepseek-ai/DeepSeek-V3-0324-fast | 73.57 tok/s | 0.62s | 5 |
deepseek/deepseek-v3 | 72.22 tok/s | 5.95s | 40 |
deepseek/deepseek-v3-0324 | 71.19 tok/s | 6.86s | 235 |
Pro/deepseek-ai/DeepSeek-V3 | 45.96 tok/s | 2.69s | 5 |
deepseek-v3-0324 | 42.71 tok/s | 3.46s | 65 |
deepseek-v3 | 40.72 tok/s | 1.24s | 20 |
deepseek-v3-0324-free | 39.08 tok/s | 2.44s | 5 |
deepseek-ai/DeepSeek-V3-0324 | 35.94 tok/s | 1.85s | 120 |
deepseek-v3 | 34.60 tok/s | 0.76s | 5 |
DeepSeek-v3:3nGcHxhAPkkDukSuJhLGUX | 34.42 tok/s | 3.40s | 5 |
DeepSeek-v3 | 32.97 tok/s | 1.18s | 70 |
DeepSeek-V3 | 32.26 tok/s | 2.19s | 30 |
DeepSeek-V3-0324-128k | 26.96 tok/s | 1.15s | 10 |
deepseek-v3 | 26.71 tok/s | 3.41s | 5 |
deepseek-v3 | 24.93 tok/s | 2.06s | 5 |
Показано 1–20 из 30 провайдеров
deepseek-r1
DeepSeek R1 is a reasoning-focused language model in the DeepSeek series, designed for complex reasoning, problem-solving, and analytical tasks.
deepseek-v3-2
DeepSeek V3.2 is an upgraded V3-series MoE model with stronger reasoning, coding, and math performance, widely available through OpenAI-compatible API relays.
gpt-5-4
OpenAI GPT-5.4 extends the GPT-5 family with stronger instruction following, deeper tool use, and improved performance on coding, math, and long-document analysis.
kimi-k2-5
Moonshot Kimi K2.5 is an open-weight multimodal agent model with native vision and text input, strong coding performance, and a 256K context window.
glm-5
Zhipu GLM-5 is Zhipu flagship GLM series model with enhanced reasoning, agent capabilities, and strong performance on Chinese enterprise and coding scenarios.
deepseek-v3-1
DeepSeek V3.1 is an open-weights style frontier model from DeepSeek with strong math, coding, and Chinese-English bilingual reasoning.
Рейтинги основаны на тестах, предоставленных сообществом, и периодических зондах работоспособности. Носит рекомендательный характер, не является официальными данными.Artificial Analysis
