大模型定价
17+ 大模型 API 价格实时对比，找到最具性价比的模型
实时数据更新于 尚未刷新
DeepSeek V4-Flash
DeepSeek
最便宜
Input / 1M
$0.14
Output / 1M
$0.28
上下文128K
极致性价比MoE 架构代码生成
直连有免费层MMLU ~88%
DeepSeek V4-Pro
DeepSeek
Input / 1M
$1.74
Output / 1M
$3.48
上下文128K
1.6T 旗舰MoE 架构推理增强
直连MMLU ~92%
DeepSeek R1
DeepSeek
Input / 1M
$0.55
Output / 1M
$2.19
上下文128K
推理增强思维链数学专精
直连MATH ~95%
GPT-4.1-nano
OpenAI
超低价
Input / 1M
$0.10
Output / 1M
$0.40
上下文1M
超低价1M 上下文指令遵循
需翻墙
GPT-4.1
OpenAI
Input / 1M
$2.00
Output / 1M
$8.00
上下文1M
1M 上下文代码专精指令遵循
需翻墙
GPT-5.5
OpenAI
旗舰
Input / 1M
$5.00
Output / 1M
$30.00
上下文256K
最新旗舰超强推理多模态
需翻墙
Claude Haiku 4.5
Anthropic
Input / 1M
$1.00
Output / 1M
$5.00
上下文200K
轻量快速200K 上下文缓存 90% 折扣
需翻墙
Claude Sonnet 4.6
Anthropic
Input / 1M
$3.00
Output / 1M
$15.00
上下文200K
代码专精均衡之选200K 上下文
需翻墙SWE-Bench ~72%
Claude Opus 4.8
Anthropic
推理最强
Input / 1M
$5.00
Output / 1M
$25.00
上下文200K
最强推理200K 上下文缓存 90% 折扣
需翻墙SWE-Bench ~82%
Gemini 2.5 Flash-Lite
超低价
Input / 1M
$0.10
Output / 1M
$0.40
上下文1M
超低价1M 上下文免费层
需翻墙AI Studio 免费
Gemini 2.5 Flash
Best Value
Input / 1M
$0.30
Output / 1M
$2.50
上下文1M
1M 上下文多模态免费层
需翻墙AI Studio 免费
Gemini 2.5 Pro
Input / 1M
$1.25
Output / 1M
$5.00
上下文1M
1M 上下文多模态旗舰视频理解
需翻墙AI Studio 有限免费MMLU ~90%
Llama 4 Scout
Meta
10M上下文
Input / 1M
$0.08
Output / 1M
$0.30
上下文10M
10M 超长上下文开源可自部署109B MoE
直连+翻墙
Llama 4 Maverick
Meta
开源旗舰
Input / 1M
$0.20
Output / 1M
$0.80
上下文1M
400B MoE开源最强1M 上下文
直连+翻墙
GLM-4-Flash
智谱 AI
永久免费
Input / 1M
免费
Output / 1M
免费
上下文128K
永久免费30 RPM中文优秀
直连永久免费，无额度上限
GLM-5
智谱 AI
Input / 1M
¥5.00
Output / 1M
¥15.00
上下文128K
智谱旗舰推理增强中文领先
直连C-Eval ~93%
Qwen-Turbo
阿里巴巴
国产超低价
Input / 1M
¥0.30
Output / 1M
¥0.60
上下文128K
超低价快速响应代码生成
直连
Qwen 3 235B
阿里巴巴
Input / 1M
¥4.00
Output / 1M
¥12.00
上下文128K
最新旗舰235B 参数代码生成
直连C-Eval ~95%
Kimi (Moonshot)
月之暗面
Input / 1M
¥2.00
Output / 1M
¥6.00
上下文128K
长上下文联网搜索文件解析
直连
MiniMax
稀宇科技
Input / 1M
¥1.00
Output / 1M
¥2.00
上下文128K
超低价多模态语音合成
直连
Mistral Large 3
Mistral AI
EU-based
Input / 1M
$2.00
Output / 1M
$6.00
上下文128K
Top-tier reasoningMultilingualTool calls
需翻墙MMLU ~81%
Codestral 25.01
Mistral AI
Code-tuned
Input / 1M
$0.30
Output / 1M
$0.90
上下文256K
Code-specialized256K context80+ languages
需翻墙HumanEval ~92%
Mistral Medium 3
Mistral AI
Input / 1M
$0.40
Output / 1M
$2.00
上下文128K
Cost-efficientEnterprise-gradeTool calls
需翻墙
Command A
Cohere
RAG leader
Input / 1M
$2.50
Output / 1M
$10.00
上下文256K
111B paramsRAG-optimized256K context
需翻墙MMLU ~85%
Command R+
Cohere
Input / 1M
$2.50
Output / 1M
$10.00
上下文128K
RAG workflowsTool use104B params
需翻墙
Command R
Cohere
Input / 1M
$0.15
Output / 1M
$0.60
上下文128K
Cost-efficient RAGTool use35B params
需翻墙
Grok 4
xAI
Frontier
Input / 1M
$3.00
Output / 1M
$15.00
上下文256K
Frontier reasoning256K contextReal-time data
需翻墙GPQA ~75%
Grok 4 Fast
xAI
Input / 1M
$0.20
Output / 1M
$1.50
上下文128K
Fast inference1ms latency tierTool use
需翻墙
Sonar Pro
Perplexity
Online + RAG
Input / 1M
$3.00
Output / 1M
$15.00
上下文200K
Online searchCitations200K context
需翻墙
Sonar Reasoning Pro
Perplexity
Input / 1M
$2.00
Output / 1M
$8.00
上下文127K
Chain-of-thoughtOnline searchCitations
需翻墙
Phi-4
Microsoft
SLM
Input / 1M
$0.07
Output / 1M
$0.14
上下文16K
14B SLMUltra-cheapStrong reasoning
直连MMLU ~80%
Phi-4 Mini
Microsoft
Cheapest
Input / 1M
$0.01
Output / 1M
$0.03
上下文128K
3.8B params128K contextMobile-ready
直连
Llama Nemotron 405B
NVIDIA
Reasoning
Input / 1M
$0.50
Output / 1M
$1.50
上下文128K
405B paramsReasoning-tunedTool use
直连+翻墙
Llama Nemotron 70B
NVIDIA
Input / 1M
$0.12
Output / 1M
$0.36
上下文128K
70B paramsReasoning-tunedTool use
直连+翻墙
Qwen 3 Plus
Alibaba
Input / 1M
$0.40
Output / 1M
$1.20
上下文128K
MultilingualTool useCost-efficient
直连+翻墙
Qwen VL Max
Alibaba
Multimodal
Input / 1M
$2.70
Output / 1M
$2.70
上下文32K
Multimodal visionOCRImage understanding
直连
Hunyuan Turbo
Tencent
Input / 1M
¥1.50
Output / 1M
¥5.00
上下文256K
256K contextMultilingualWeChat ecosystem
直连
Kimi K2
Moonshot AI
Agentic
Input / 1M
$0.60
Output / 1M
$2.50
上下文200K
1T MoEAgentic tool use200K context
直连SWE-Bench ~70%
Pangu Ultra
Huawei
Input / 1M
¥8.00
Output / 1M
¥24.00
上下文32K
Huawei Cloud onlyEnterprise-gradeIndustry-tuned
直连
ERNIE 5.0
Baidu
Frontier
Input / 1M
¥4.00
Output / 1M
¥12.00
上下文128K
Frontier Chinese LLMMultimodalTool use
直连
ERNIE Speed
Baidu
Input / 1M
¥0.40
Output / 1M
¥0.80
上下文128K
Cost-efficientFast inference128K context
直连
Doubao Pro
ByteDance
Input / 1M
¥0.80
Output / 1M
¥2.00
上下文256K
256K contextMultimodalVolcano engine
直连
Doubao Lite
ByteDance
Ultra-cheap
Input / 1M
¥0.30
Output / 1M
¥0.60
上下文128K
Ultra-cheap128K contextFast inference
直连
Step-2
StepFun
Input / 1M
¥3.80
Output / 1M
¥13.00
上下文32K
1T MoEMultimodalVideo understanding
直连
Yi Large
01.AI
Input / 1M
$2.50
Output / 1M
$2.50
上下文32K
Bilingual EN/CNTool use32K context
直连+翻墙
DeepSeek V3.5 (Together)
Together AI
Input / 1M
$0.30
Output / 1M
$0.60
上下文128K
Hosted DeepSeekLow latency US/EUOpenAI-compatible API
需翻墙
Qwen 3 (Together)
Together AI
Input / 1M
$0.20
Output / 1M
$0.40
上下文128K
Hosted QwenLow latency US/EUOpenAI-compatible API
需翻墙
DeepSeek V3.5 (Fireworks)
Fireworks AI
Input / 1M
$0.30
Output / 1M
$0.90
上下文128K
Hosted DeepSeekFast US inferenceOpenAI-compatible API
需翻墙
DeepSeek V3.5 (Novita)
Novita AI
Cheapest
Input / 1M
$0.20
Output / 1M
$0.50
上下文128K
Cheapest DeepSeek hostGlobal routingOpenAI-compatible API
直连
OpenRouter Auto
OpenRouter
Aggregator
Input / 1M
~$0.30
Output / 1M
~$0.90
上下文128K
Auto-routing300+ modelsSingle API key
需翻墙
模型横向对比
一览所有模型的核心参数与定价，快速找到最适合你的模型
| 模型 | 提供商 | Input Price | Output Price | 上下文 | 免费层 | 访问 | Benchmark |
|---|---|---|---|---|---|---|---|
DeepSeek V4-Flash最便宜 | DeepSeek | $0.14 | $0.28 | 128K | 有免费层 | 直连 | MMLU ~88% |
DeepSeek V4-Pro | DeepSeek | $1.74 | $3.48 | 128K | — | 直连 | MMLU ~92% |
DeepSeek R1 | DeepSeek | $0.55 | $2.19 | 128K | — | 直连 | MATH ~95% |
GPT-4.1-nano超低价 | OpenAI | $0.10 | $0.40 | 1M | — | 需翻墙 | — |
GPT-4.1 | OpenAI | $2.00 | $8.00 | 1M | — | 需翻墙 | — |
GPT-5.5旗舰 | OpenAI | $5.00 | $30.00 | 256K | — | 需翻墙 | — |
Claude Haiku 4.5 | Anthropic | $1.00 | $5.00 | 200K | — | 需翻墙 | — |
Claude Sonnet 4.6 | Anthropic | $3.00 | $15.00 | 200K | — | 需翻墙 | SWE-Bench ~72% |
Claude Opus 4.8推理最强 | Anthropic | $5.00 | $25.00 | 200K | — | 需翻墙 | SWE-Bench ~82% |
Gemini 2.5 Flash-Lite超低价 | $0.10 | $0.40 | 1M | AI Studio 免费 | 需翻墙 | — | |
Gemini 2.5 FlashBest Value | $0.30 | $2.50 | 1M | AI Studio 免费 | 需翻墙 | — | |
Gemini 2.5 Pro | $1.25 | $5.00 | 1M | AI Studio 有限免费 | 需翻墙 | MMLU ~90% | |
Llama 4 Scout10M上下文 | Meta | $0.08 | $0.30 | 10M | — | 直连+翻墙 | — |
Llama 4 Maverick开源旗舰 | Meta | $0.20 | $0.80 | 1M | — | 直连+翻墙 | — |
GLM-4-Flash永久免费 | 智谱 AI | 免费 | 免费 | 128K | 永久免费，无额度上限 | 直连 | — |
GLM-5 | 智谱 AI | ¥5.00 | ¥15.00 | 128K | — | 直连 | C-Eval ~93% |
Qwen-Turbo国产超低价 | 阿里巴巴 | ¥0.30 | ¥0.60 | 128K | — | 直连 | — |
Qwen 3 235B | 阿里巴巴 | ¥4.00 | ¥12.00 | 128K | — | 直连 | C-Eval ~95% |
Kimi (Moonshot) | 月之暗面 | ¥2.00 | ¥6.00 | 128K | — | 直连 | — |
MiniMax | 稀宇科技 | ¥1.00 | ¥2.00 | 128K | — | 直连 | — |
Mistral Large 3EU-based | Mistral AI | $2.00 | $6.00 | 128K | — | 需翻墙 | MMLU ~81% |
Codestral 25.01Code-tuned | Mistral AI | $0.30 | $0.90 | 256K | — | 需翻墙 | HumanEval ~92% |
Mistral Medium 3 | Mistral AI | $0.40 | $2.00 | 128K | — | 需翻墙 | — |
Command ARAG leader | Cohere | $2.50 | $10.00 | 256K | — | 需翻墙 | MMLU ~85% |
Command R+ | Cohere | $2.50 | $10.00 | 128K | — | 需翻墙 | — |
Command R | Cohere | $0.15 | $0.60 | 128K | — | 需翻墙 | — |
Grok 4Frontier | xAI | $3.00 | $15.00 | 256K | — | 需翻墙 | GPQA ~75% |
Grok 4 Fast | xAI | $0.20 | $1.50 | 128K | — | 需翻墙 | — |
Sonar ProOnline + RAG | Perplexity | $3.00 | $15.00 | 200K | — | 需翻墙 | — |
Sonar Reasoning Pro | Perplexity | $2.00 | $8.00 | 127K | — | 需翻墙 | — |
Phi-4SLM | Microsoft | $0.07 | $0.14 | 16K | — | 直连 | MMLU ~80% |
Phi-4 MiniCheapest | Microsoft | $0.01 | $0.03 | 128K | — | 直连 | — |
Llama Nemotron 405BReasoning | NVIDIA | $0.50 | $1.50 | 128K | — | 直连+翻墙 | — |
Llama Nemotron 70B | NVIDIA | $0.12 | $0.36 | 128K | — | 直连+翻墙 | — |
Qwen 3 Plus | Alibaba | $0.40 | $1.20 | 128K | — | 直连+翻墙 | — |
Qwen VL MaxMultimodal | Alibaba | $2.70 | $2.70 | 32K | — | 直连 | — |
Hunyuan Turbo | Tencent | ¥1.50 | ¥5.00 | 256K | — | 直连 | — |
Kimi K2Agentic | Moonshot AI | $0.60 | $2.50 | 200K | — | 直连 | SWE-Bench ~70% |
Pangu Ultra | Huawei | ¥8.00 | ¥24.00 | 32K | — | 直连 | — |
ERNIE 5.0Frontier | Baidu | ¥4.00 | ¥12.00 | 128K | — | 直连 | — |
ERNIE Speed | Baidu | ¥0.40 | ¥0.80 | 128K | — | 直连 | — |
Doubao Pro | ByteDance | ¥0.80 | ¥2.00 | 256K | — | 直连 | — |
Doubao LiteUltra-cheap | ByteDance | ¥0.30 | ¥0.60 | 128K | — | 直连 | — |
Step-2 | StepFun | ¥3.80 | ¥13.00 | 32K | — | 直连 | — |
Yi Large | 01.AI | $2.50 | $2.50 | 32K | — | 直连+翻墙 | — |
DeepSeek V3.5 (Together) | Together AI | $0.30 | $0.60 | 128K | — | 需翻墙 | — |
Qwen 3 (Together) | Together AI | $0.20 | $0.40 | 128K | — | 需翻墙 | — |
DeepSeek V3.5 (Fireworks) | Fireworks AI | $0.30 | $0.90 | 128K | — | 需翻墙 | — |
DeepSeek V3.5 (Novita)Cheapest | Novita AI | $0.20 | $0.50 | 128K | — | 直连 | — |
OpenRouter AutoAggregator | OpenRouter | ~$0.30 | ~$0.90 | 128K | — | 需翻墙 | — |