78 lines
1.9 KiB
Markdown
78 lines
1.9 KiB
Markdown
---
|
||
title: OpenRouter 模型性价比分析
|
||
source:
|
||
author: shenwei
|
||
published:
|
||
created:
|
||
description:
|
||
tags: []
|
||
---
|
||
|
||
# OpenRouter 模型性价比分析
|
||
**来源**: https://openrouter.ai/models
|
||
**日期**: 2026-04-03
|
||
**总模型数**: 349（免费: 28，付费: 319）
|
||
|
||
---
|
||
|
||
## 免费模型
|
||
|
||
| 模型 | 上下文 | 亮点 |
|
||
| ----------------------- | ---- | ------------------- |
|
||
| Qwen3.6 Plus (free) | 1M | 最新款，SWE-bench 78.8% |
|
||
| NVIDIA Nemotron 3 Super | 262K | 120B MoE 仅激活12B |
|
||
| MiniMax M2.5 (free) | 196K | 多Agent协同 |
|
||
| StepFun 3.5 Flash | 256K | 国产开源 |
|
||
| Arcee AI: Trinity Large | 131K | 推理专用，前5天免费 |
|
||
|
||
---
|
||
|
||
## 付费性价比 TOP 5
|
||
|
||
| 排名 | 模型 | 每1K成本 | 上下文 |
|
||
|------|------|---------|--------|
|
||
| 1 | Qwen3.5-Flash | $0.00032 | 1M |
|
||
| 2 | Grok 4.1 Fast | $0.00070 | 2M |
|
||
| 3 | Gemini 2.0 Flash Lite | $0.00040 | 1M |
|
||
| 4 | Gemini 2.5 Flash Lite | $0.00050 | 1M |
|
||
| 5 | Qwen3.5-9B | $0.00020 | 256K |
|
||
|
||
---
|
||
|
||
## 编程/Coding 专项
|
||
|
||
| 模型 | 每1K成本 | 上下文 |
|
||
| -------------------------- | -------- | ---- |
|
||
| Qwen2.5 Coder 7B | $0.00012 | 32K |
|
||
| Mistral Small 3.1 | $0.00014 | 131K |
|
||
| Qwen3 235B A22B | $0.00017 | 262K |
|
||
| Kwaipilot KAT-Coder-Pro V2 | $0.00150 | 256K |
|
||
|
||
---
|
||
|
||
## 旗舰模型
|
||
|
||
| 模型 | 每1K成本 | 上下文 |
|
||
|------|---------|--------|
|
||
| GPT-5.4 Pro | $0.00021 | 1M+ |
|
||
| GPT-5.4 | $0.00043 | 1M+ |
|
||
| Grok 4.20 | $0.00080 | 2M |
|
||
| MiniMax M2.7 | $0.00050 | 204K |
|
||
|
||
---
|
||
|
||
## 星枢推荐组合
|
||
|
||
```
|
||
日常轻量: Qwen3.5-Flash ($0.00032/1K)
|
||
编程任务: Qwen3 235B A22B / Qwen2.5 Coder 7B
|
||
长上下文: Grok 4.1 Fast / Gemini 2.0 Flash Lite
|
||
免费备用: Qwen3.6 Plus / MiniMax M2.5
|
||
旗舰需求: GPT-5.4 Pro / Grok 4.20
|
||
```
|
||
|
||
---
|
||
|
||
## 原始数据
|
||
- `openrouter-models-2026-04-03.json` (484KB)
|