免費 / 開源由 DeepSeek-V4 與 R1 驅動的高效開源權重與託管推理 AI,token 成本極低。

依功能與用途,為你挑選 12 款 General Reasoning 的替代工具。
| 排名 | 工具與標語 | 官方網域 | 每月造訪次數 | 月增幅 |
|---|
| 定價模式 |
|---|
| 主要市場 |
|---|
| 評分 |
|---|
| 比較 |
|---|
| ● | General Reasoning | gr.inc | 9441 | 3838 (+68.5%) | 免費 | 全球 | 4.2 | 簡介 |
| 1 | DeepSeek | deepseek.com | 3.73億 | 1536.22萬 (+4.3%) | 免費增值 | 中國 | 4.8 | 比較 |
| 2 | Google Gemma 開源模型 | ai.google.dev | 490.09萬 | -76.28萬 (-13.5%) | 免費增值、付費 | 美國 | 4.8 | 比較 |
| 3 | ModelScope | modelscope.cn | 308.54萬 | 14.91萬 (+5.1%) | 免費 | 中國 | 4.5 | 比較 |
| 4 | OpenAI 平台 | openai.com | 2.35億 | -2478.28萬 (-9.6%) | 付費 | 美國 | 4.8 | 比較 |
| 5 | Qwen 4 (通義千問 4) | qwen.ai | 3510.14萬 | -233.98萬 (-6.2%) | 免費增值 | 俄羅斯 | 4.7 | 比較 |
| 6 | Gemini 4 Carbon (Gemini Carbon 4) | deepmind.google | 433.94萬 | 48.98萬 (+12.7%) | 免費增值 | 美國 | 4.8 | 比較 |
| 7 | Civitai | civitai.com | 696.31萬 | 79.2萬 (+12.8%) | 免費增值 | 美國 | 4.6 | 比較 |
| 8 | DeepInfra | deepinfra.com | 45.78萬 | -4.45萬 (-8.9%) | 付費 | 美國 | 4.6 | 比較 |
| 9 | Together AI | together.ai | 164.59萬 | 64.77萬 (+64.9%) | 免費增值、付費 | 美國 | 4.7 | 比較 |
| 10 | Unsloth AI | unsloth.ai | 88.75萬 | -19.01萬 (-17.6%) | 免費 | 美國 | 4.8 | 比較 |
| 11 | AI/ML API | aimlapi.com | 23.29萬 | -5.87萬 (-20.1%) | Pay-as-you-go credits (free to start); enterprise custom pricing on request | 美國 | 4.3 | 比較 |
| 12 | TypeSafe AI - Jev | typesafe.ai | 614.87萬 | -351.3萬 (-36.4%) | 付費 | 美國 | 4.2 | 比較 |
免費 / 開源由 DeepSeek-V4 與 R1 驅動的高效開源權重與託管推理 AI,token 成本極低。
免費 / 開源ModelScope (魔搭社区) is an open-source AI community and model service platform that hosts thousands of pretrained models, datasets, and ready-to-run demos, backed by Alibaba Cloud infrastructure.
免費 / 開源ModelScope (魔搭社区) is an open-source AI community and model service platform that hosts thousands of pretrained models, datasets, and ready-to-run demos, backed by Alibaba Cloud infrastructure.

免費 / 開源阿里雲第四代基礎大模型家族與全模態 AI 助手 — 涵蓋 Qwen 4 (Qwen4)、Qwen4-Max、Qwen4-Plus、Qwen4-Flash、Qwen4-27B 與 Qwen 3.8-Flash-Next 架構
免費 / 開源阿里雲第四代基礎大模型家族與全模態 AI 助手 — 涵蓋 Qwen 4 (Qwen4)、Qwen4-Max、Qwen4-Plus、Qwen4-Flash、Qwen4-27B 與 Qwen 3.8-Flash-Next 架構

專項更強Google DeepMind 榜首級資安防禦與長週期智慧體工程前沿模型 — Gemini 4 Carbon (Gemini Carbon 4),繼 Gemini 4 Argon 後的巔峰之作
專項更強Google DeepMind 榜首級資安防禦與長週期智慧體工程前沿模型 — Gemini 4 Carbon (Gemini Carbon 4),繼 Gemini 4 Argon 後的巔峰之作

Together AI is the AI Native Cloud — a full-stack platform for running, fine-tuning and training open-source models. It spans serverless inference APIs, dedicated endpoints, GPU clusters, post-training and storage, all backed by in-house systems research.
Together AI is the AI Native Cloud — a full-stack platform for running, fine-tuning and training open-source models. It spans serverless inference APIs, dedicated endpoints, GPU clusters, post-training and storage, all backed by in-house systems research.
免費 / 開源Unsloth is an open-source toolkit for running and training large language models locally. It provides faster, more memory-efficient fine-tuning, a local chat and agent environment with browser use, image and video generation, and an OpenAI-compatible API your existing tools can connect to.
免費 / 開源Unsloth is an open-source toolkit for running and training large language models locally. It provides faster, more memory-efficient fine-tuning, a local chat and agent environment with browser use, image and video generation, and an OpenAI-compatible API your existing tools can connect to.

一個通用 API,串接 500+ AI 模型與 1,000+ 代理工具

專項更強TypeSafe AI is an AI lab building machine-native intelligence infrastructure, and Jev is its first System One Model — a model designed to make decisions inside software rather than reply to humans. Instead of returning strings, Jev accepts structured questions and returns typed decisions with calibrated confidence scores, so applications can act autonomously when confidence is high and escalate for human review when it is not. TypeSafe reports roughly 193x faster and 444x cheaper performance on System One workloads compared with general-purpose LLMs.
專項更強TypeSafe AI is an AI lab building machine-native intelligence infrastructure, and Jev is its first System One Model — a model designed to make decisions inside software rather than reply to humans. Instead of returning strings, Jev accepts structured questions and returns typed decisions with calibrated confidence scores, so applications can act autonomously when confidence is high and escalate for human review when it is not. TypeSafe reports roughly 193x faster and 444x cheaper performance on System One workloads compared with general-purpose LLMs.