Mistral AI 免费 API 详情
实测于 2026-08-23 00:30 +08:00（来源：oneapi 生产 DB Channel 表直读）
渠道 slug： | 详情页 URL：
1. 渠道介绍
Mistral AI 是一家领先的欧洲 AI 公司，免费层包含 35 个对话模型和 4 个 Voxtral 音频模型。组合覆盖代码（Codestral）、推理（Magistral）、效率（Ministral）和通用对话（Mistral Large/Medium/Small）。60 RPM 限制，免费层中最高。
2. 实测支持模型列表
口径：models_total = 39（Channel.models 全量直读）；chat = 35；audio = 4；embedding = 0；moderation = 0
对话模型（35 个）
- codestral-2508
- codestral-latest
- devstral-2512
- devstral-latest
- devstral-medium-latest
- glm-5-2
- labs-leanstral-1-5
- labs-leanstral-1-5-1
- magistral-medium-latest
- magistral-small-latest
- ministral-14b-2512
- ministral-14b-latest
- ministral-3b-2512
- ministral-3b-latest
- ministral-8b-2512
- ministral-8b-latest
- mistral-code-agent-latest
- mistral-code-fim-latest
- mistral-code-latest
- mistral-large-2512
- mistral-large-latest
- mistral-medium
- mistral-medium-2505
- mistral-medium-2508
- mistral-medium-2604
- mistral-medium-3
- mistral-medium-3-5
- mistral-medium-3.5
- mistral-medium-latest
- mistral-small-2603
- mistral-small-latest
- mistral-vibe-cli-fast
- mistral-vibe-cli-latest
- mistral-vibe-cli-with-tools
- zai-glm-5-2
音频模型（4 个）
- voxtral-mini-2602
- voxtral-mini-latest
- voxtral-small-2507
- voxtral-small-latest
3. 注册方法
待定
4. 使用方法
Base URL:
待定
5. 实测数据
待 it 基准实测 JSON 到料后填充。每数据点标注「实测于 YYYY-MM-DD HH:MM」。
| 指标 |
值 |
备注 |
| HTTP 状态码 |
待填 |
实测 /models 端点 |
| 首 token 延迟 |
待填 |
ms |
| 完整响应时间 |
待填 |
ms |
| TPS |
待填 |
tokens/sec |
| RPM 限制 |
待填 |
429 实测门槛 |
| TPM 限制 |
待填 |
429 实测门槛 |
| 日调用上限 |
待填 |
429 实测门槛 |
6. 优缺点总结
优点：
- 35 个对话模型覆盖所有用例
- 4 个 Voxtral 音频模型
- 60 RPM 免费层限制最高
缺点：
- 大量 -latest 变体可能令人困惑
- 音频模型为专用型
本页由内容维护员于 2026-08-25 生成，数据以 oneapi 权威 JSON 为准。
Mistral AI Free API Details
Verified on 2026-08-23 00:30 +08:00 (source: OneAPI production DB Channel table direct read)
Channel slug: | Detail URL:
1. Channel Overview
Mistral AI is a leading European AI company with 35 chat models and 4 Voxtral audio models in the free tier. The portfolio covers code (Codestral), reasoning (Magistral), efficiency (Ministral), and general chat (Mistral Large/Medium/Small). 60 RPM limit, the highest among free tiers.
2. Verified Supported Models
Scope: models_total = 39 (full Channel.models direct read); chat = 35; audio = 4; embedding = 0; moderation = 0
Chat Models (35)
- codestral-2508
- codestral-latest
- devstral-2512
- devstral-latest
- devstral-medium-latest
- glm-5-2
- labs-leanstral-1-5
- labs-leanstral-1-5-1
- magistral-medium-latest
- magistral-small-latest
- ministral-14b-2512
- ministral-14b-latest
- ministral-3b-2512
- ministral-3b-latest
- ministral-8b-2512
- ministral-8b-latest
- mistral-code-agent-latest
- mistral-code-fim-latest
- mistral-code-latest
- mistral-large-2512
- mistral-large-latest
- mistral-medium
- mistral-medium-2505
- mistral-medium-2508
- mistral-medium-2604
- mistral-medium-3
- mistral-medium-3-5
- mistral-medium-3.5
- mistral-medium-latest
- mistral-small-2603
- mistral-small-latest
- mistral-vibe-cli-fast
- mistral-vibe-cli-latest
- mistral-vibe-cli-with-tools
- zai-glm-5-2
Audio Models (4)
- voxtral-mini-2602
- voxtral-mini-latest
- voxtral-small-2507
- voxtral-small-latest
3. Registration
TBD
4. Usage
Base URL:
TBD
5. Verified Metrics
待 it 基准实测 JSON 到料后填充。每数据点标注「实测于 YYYY-MM-DD HH:MM」。
| Metric |
Value |
Notes |
| HTTP Status |
TBD |
via /models endpoint |
| Time to First Token |
TBD |
ms |
| Total Response Time |
TBD |
ms |
| TPS |
TBD |
tokens/sec |
| RPM Limit |
TBD |
429 threshold (probed) |
| TPM Limit |
TBD |
429 threshold (probed) |
| Daily Limit |
TBD |
429 threshold (probed) |
6. Pros & Cons
Pros:
- 35 chat models for every use case
- 4 Voxtral audio models
- Highest 60 RPM free tier limit
Cons:
- Many -latest variants can be confusing
- Audio models are specialized
本页由内容维护员于 2026-08-25 生成，数据以 oneapi 权威 JSON 为准。