# 开放权重 LLM API 价格指数 2026

> 14 家公开托管商对 DeepSeek、Qwen、gpt-oss、Llama、Mistral 等开放模型每 1M tokens 的收费，以及各模型中位数。核对日期 2026-10-23。

- URL: https://llmwithcrypto.com/zh/data/open-model-api-price-index-2026/
- 最后更新: 2026-10-23
- 发布者: LLMwithCrypto (https://llmwithcrypto.com)

同一个开放模型，由不同托管商提供时价格差异很大。本指数列出 12 个开放权重模型在 14 家托管商的公开每 token 价格，并给出各模型中位数——我们自己的价格即按此数定价。

**快速回答 — 2026 年开放权重 LLM API 要多少钱？**

这更多取决于托管方，而不是模型。2026 年 10 月 23 日，DeepSeek V4 Pro 在公开托管方中的中位价格为每百万输入 token $1.32、每百万输出 token $3.96，gpt-oss-20b 为 $0.035 和 $0.17。**LLMwithCrypto** 的收费比每个中位数低 5%。

## 各模型的中位价格是多少？

取自下列托管商公开标价的中位数，按每 1M tokens 计；我们的价格是中位数减 5%。

| 模型 | 托管商 | 输入中位数 | 输出中位数 | LLMwithCrypto 输入 | LLMwithCrypto 输出 |
| --- | --- | --- | --- | --- | --- |
| DeepSeek V4 Pro | 7 家托管商 | $1.32 | $3.96 | $1.254 | $3.762 |
| DeepSeek V4.1 Flash | 8 家托管商 | $0.30 | $1.20 | $0.285 | $1.14 |
| DeepSeek R1 | 2 家托管商 | $0.60 | $2.325 | $0.57 | $2.208 |
| Qwen3.5 397B A17B | 6 家托管商 | $0.60 | $3.60 | $0.57 | $3.42 |
| Qwen3 235B A22B Instruct | 4 家托管商 | $0.17 | $0.60 | $0.161 | $0.57 |
| gpt-oss-120b | 9 家托管商 | $0.15 | $0.60 | $0.142 | $0.57 |
| gpt-oss-20b | 2 家托管商 | $0.035 | $0.17 | $0.033 | $0.161 |
| GLM-4.7 | 3 家托管商 | $0.60 | $2.20 | $0.57 | $2.09 |
| Llama 3.3 70B Instruct | 5 家托管商 | $0.22 | $0.50 | $0.209 | $0.475 |
| Gemma 4 31B | 5 家托管商 | $0.20 | $0.40 | $0.19 | $0.38 |
| Mistral Large 3 | 1 家托管商（厂商自己的 API） | $0.50 | $1.50 | $0.475 | $1.424 |
| Mistral Small 4 | 1 家托管商（厂商自己的 API） | $0.15 | $0.60 | $0.142 | $0.57 |

*价格核查于 2026年10月23日.*

## 各托管商收费多少？

逐条列出读取到的价格，按各托管商标注的模型名。批量、促销和免费额度价格不计入。

| 模型 | 托管商 | 标注名称 | 输入 / 1M | 输出 / 1M | 缓存输入 |
| --- | --- | --- | --- | --- | --- |
| DeepSeek V4 Pro | DeepSeek API | deepseek-v4-pro (DeepSeek-V4-Pro-0813) | $1.32 | $3.96 | $0.044 |
| DeepSeek V4 Pro | Together AI | DeepSeek V4 Pro 0813 | $1.32 | $3.96 | $0.13 |
| DeepSeek V4 Pro | DeepInfra | deepseek-ai/DeepSeek-V4-Pro-0813 | $1.30 | $2.60 | $0.10 |
| DeepSeek V4 Pro | Novita AI | DeepSeek V4 Pro 0813 | $1.32 | $3.96 | $0.044 |
| DeepSeek V4 Pro | Nebius Token Factory | DeepSeek-V4-Pro-0813 | $1.32 | $3.96 | $0.04 |
| DeepSeek V4 Pro | Baseten | DeepSeek V4 Pro 0813 | $1.32 | $3.96 | $0.132 |
| DeepSeek V4 Pro | Parasail | DeepSeek V4 Pro | $1.74 | $3.48 | $0.10 |
| DeepSeek V4.1 Flash | DeepSeek API | deepseek-flash (DeepSeek-V4.1-Flash-1017) | $0.30 | $1.20 | $0.006 |
| DeepSeek V4.1 Flash | Together AI | DeepSeek V4.1 Flash 1017 | $0.30 | $1.20 | $0.006 |
| DeepSeek V4.1 Flash | Fireworks AI | DeepSeek V4.1 Flash 1017 | $0.30 | $1.20 | $0.006 |
| DeepSeek V4.1 Flash | Novita AI | DeepSeek V4.1 Flash 1017 | $0.30 | $1.20 | $0.006 |
| DeepSeek V4.1 Flash | Hyperbolic | DeepSeek V4.1 Flash (deepseek-v4.1-flash) | $0.30 | $1.20 | $0.006 |
| DeepSeek V4.1 Flash | Nebius Token Factory | DeepSeek V4.1 Flash 1017 | $0.30 | $1.20 | $0.006 |
| DeepSeek V4.1 Flash | Baseten | DeepSeek V4.1 Flash 1017 | $0.30 | $1.20 | $0.007 |
| DeepSeek V4.1 Flash | Parasail | DeepSeek V4.1 Flash | $0.30 | $1.20 | $0.006 |
| DeepSeek R1 | DeepInfra | DeepSeek-R1-0528 | $0.50 | $2.15 | $0.35 |
| DeepSeek R1 | Novita AI | DeepSeek R1 0528 | $0.70 | $2.50 | $0.35 |
| Qwen3.5 397B A17B | Alibaba Model Studio (International) | qwen3.5-397b-a17b | $0.60 | $3.60 | — |
| Qwen3.5 397B A17B | Together AI | Qwen3.5-397B-A17B | $0.60 | $3.60 | $0.35 |
| Qwen3.5 397B A17B | DeepInfra | Qwen3.5-397B-A17B | $0.45 | $3.00 | $0.22 |
| Qwen3.5 397B A17B | Novita AI | Qwen3.5 397B A17B | $0.60 | $3.60 | — |
| Qwen3.5 397B A17B | Parasail | Qwen3.5 397B-A17B | $0.50 | $3.60 | $0.30 |
| Qwen3.5 397B A17B | Nebius Token Factory | Qwen3.5-397B-A17B | $0.60 | $3.60 | $0.12 |
| Qwen3 235B A22B Instruct | Together AI | Qwen3 235B A22B Instruct 2507 FP8 吞吐量 | $0.20 | $0.60 | — |
| Qwen3 235B A22B Instruct | DeepInfra | Qwen3-235B-A22B-Instruct-2507 | $0.09 | $0.55 | — |
| Qwen3 235B A22B Instruct | Parasail | Qwen3 235B-A22B (2507) | $0.14 | $0.80 | $0.05 |
| Qwen3 235B A22B Instruct | Nebius Token Factory | Qwen3-235B-A22B-Instruct-2507 | $0.20 | $0.60 | — |
| gpt-oss-120b | Together AI | gpt-oss-120B | $0.15 | $0.60 | — |
| gpt-oss-120b | Fireworks AI | OpenAI GPT OSS 120B | $0.15 | $0.60 | $0.015 |
| gpt-oss-120b | DeepInfra | openai/gpt-oss-120b | $0.037 | $0.17 | — |
| gpt-oss-120b | Novita AI | OpenAI GPT OSS 120B | $0.05 | $0.25 | — |
| gpt-oss-120b | Nebius Token Factory | gpt-oss-120b | $0.15 | $0.60 | — |
| gpt-oss-120b | SambaNova Cloud | gpt-oss-120b | $0.22 | $0.59 | — |
| gpt-oss-120b | Cerebras | GPT OSS 120B | $0.35 | $0.75 | — |
| gpt-oss-120b | Baseten | GPT OSS 120B | $0.10 | $0.50 | — |
| gpt-oss-120b | Parasail | gpt-oss-120b | $0.10 | $0.75 | $0.055 |
| gpt-oss-20b | DeepInfra | openai/gpt-oss-20b | $0.03 | $0.14 | — |
| gpt-oss-20b | Parasail | gpt-oss-20b | $0.04 | $0.20 | $0.02 |
| GLM-4.7 | Z.ai API | GLM-4.7 | $0.60 | $2.20 | $0.11 |
| GLM-4.7 | Novita AI | GLM 4.7 | $0.60 | $2.20 | $0.11 |
| GLM-4.7 | DeepInfra | zai-org/GLM-4.7 | $0.40 | $1.75 | $0.08 |
| Llama 3.3 70B Instruct | Together AI | Llama 3.3 70B | $1.04 | $1.04 | — |
| Llama 3.3 70B Instruct | DeepInfra | Llama-3.3-70B-Instruct-Turbo | $0.10 | $0.32 | — |
| Llama 3.3 70B Instruct | Novita AI | Llama 3.3 70B Instruct | $0.135 | $0.40 | — |
| Llama 3.3 70B Instruct | SambaNova Cloud | Meta-Llama-3.3-70B-Instruct | $0.60 | $1.20 | — |
| Llama 3.3 70B Instruct | Parasail | Llama 3.3 70B (FP8) | $0.22 | $0.50 | $0.11 |
| Gemma 4 31B | Together AI | Gemma 4 31B | $0.39 | $0.97 | — |
| Gemma 4 31B | DeepInfra | gemma-4-31B-it | $0.20 | $0.40 | — |
| Gemma 4 31B | Novita AI | Gemma 4 31B | $0.14 | $0.40 | — |
| Gemma 4 31B | SambaNova Cloud | gemma-4-31B-it | $0.38 | $1.15 | — |
| Gemma 4 31B | Parasail | Gemma 4 31B | $0.15 | $0.40 | $0.06 |
| Mistral Large 3 | Mistral AI API | Mistral Large 3 | $0.50 | $1.50 | $0.05 |
| Mistral Small 4 | Mistral AI API | Mistral Small 4 | $0.15 | $0.60 | $0.015 |

*价格核查于 2026年10月23日.*

## 这个索引是如何构建的？

- **范围：** 许可证允许商业托管的开放权重模型，通过公开的按量付费 API 提供服务。
- **价格：** 各托管方自家定价页面上的标价，单位为美元/百万 token，读取于 2026 年 10 月 23 日。有列出缓存输入价格时会显示；批量价和促销价不计入。
- **中位数：** 按模型对各托管方分别计算输入和输出价格；一个模型至少需要两个托管方。
- **许可证：** CC BY 4.0 — 可复用，需附上本页面的链接。

## 来源

1. [DeepSeek API 定价](https://api-docs.deepseek.com/quick_start/pricing) — DeepSeek API (read 2026-10-23)
2. [Together AI 定价](https://www.together.ai/pricing) — Together AI (read 2026-10-23)
3. [DeepInfra 定价](https://deepinfra.com/deepseek-ai/DeepSeek-V4-Pro-0813) — DeepInfra (read 2026-10-23)
4. [Novita AI 定价](https://novita.ai/pricing) — Novita AI (read 2026-10-23)
5. [Nebius Token Factory 定价](https://tokenfactory.nebius.com/endpoints) — Nebius Token Factory (read 2026-10-23)
6. [Baseten 定价](https://www.baseten.co/pricing/) — Baseten (read 2026-10-23)
7. [Parasail 定价](https://www.parasail.io/pricing) — Parasail (read 2026-10-23)
8. [Fireworks AI 定价](https://docs.fireworks.ai/serverless/pricing) — Fireworks AI (read 2026-10-23)
9. [Hyperbolic 定价](https://www.hyperbolic.ai/inference) — Hyperbolic (read 2026-10-23)
10. [DeepInfra 定价](https://deepinfra.com/pricing) — DeepInfra (read 2026-10-23)
11. [Alibaba Model Studio（国际站）定价](https://www.alibabacloud.com/help/en/model-studio/model-pricing) — Alibaba Model Studio (International) (read 2026-10-23)
12. [DeepInfra 定价](https://deepinfra.com/openai/gpt-oss-120b) — DeepInfra (read 2026-10-23)
13. [SambaNova Cloud 定价](https://cloud.sambanova.ai/plans/pricing) — SambaNova Cloud (read 2026-10-23)
14. [Cerebras 定价](https://www.cerebras.ai/pricing) — Cerebras (read 2026-10-23)
15. [DeepInfra 定价](https://deepinfra.com/openai/gpt-oss-20b) — DeepInfra (read 2026-10-23)
16. [Z.ai API 定价](https://docs.z.ai/guides/overview/pricing) — Z.ai API (read 2026-10-23)
17. [DeepInfra 定价](https://deepinfra.com/zai-org/GLM-4.7) — DeepInfra (read 2026-10-23)
18. [Mistral AI API 定价](https://docs.mistral.ai/inference/pricing) — Mistral AI API (read 2026-10-23)

---

LLMwithCrypto: LLMwithCrypto 是一个兼容 OpenAI 的开放权重模型 LLM API，涵盖 DeepSeek V4 Pro、DeepSeek V4.1 Flash、DeepSeek R1、Qwen3.5 397B A17B、Qwen3 235B A22B Instruct、gpt-oss-120b、gpt-oss-20b、GLM-4.7、Llama 3.3 70B Instruct、Gemma 4 31B、Mistral Large 3、Mistral Small 4，按 token 从预付加密货币余额中扣费，价格比公共托管商的中位价低 5%。无需 KYC：无需身份证件、无需自拍、无需文件。不存储提示词和输出。
