Strong for its size
0.551 on Arena Hard 和 68.5 on WildBench in Mistral's tests, ahead of Qwen3 14B 和 Gemma 3 12B, and comparable to Mistral Small 3.2 24B by Mistral's account.
Mistral AI
运行正常· 每隔几分钟检查一次Ministral 3 14B API
The Ministral 3 14B API: the largest Ministral 3 model, with vision 和 tool calling, 70% below Mistral's 原价
每 1M token 价格70%优惠
- 输入
- $0.06
$0.20 - 输出
- $0.06
$0.20
模型 ID
ministral-3-14bchat.py
from openai import OpenAIclient = OpenAI( base_url="https://api.zurelay.com/v1", api_key="YOUR_ZURELAY_KEY",)stream = client.chat.completions.create( model="ministral-3-14b", messages=[{"role": "user", "content": "Explain quicksort in two sentences."}], stream=True,)for chunk in stream: if chunk.choices: print(chunk.choices[0].delta.content or "", end="")Base URL https://api.zurelay.com/v1. 支持任意 OpenAI SDK。 Chat API 文档
- 上下文窗口
- 128K tokens
- 推理
- 无
- 输入
- Text, Image
- 发布日期
- 2025年12月2日
- 状态
- 运行正常
价格
Ministral 3 14B API 价格。 仅需官方定价的一小部分。
从预付余额中按量扣费，无需订阅。$10 起充。请求失败不收费。
| 单价 | Zurelay | Mistral AI 原价 | 节省金额 |
|---|---|---|---|
输入 每 1M tokens | $0.06 | 70%优惠 | |
输出 每 1M tokens | $0.06 | 70%优惠 |
所有提示词长度均采用同一费率, up to the full 128K context window。流式输出、工具调用和结构化输出不额外收费。
省钱计算器
拖动滑块设置每月用量。
每月输入 token 数50M
每月输出 token 数10M
Zurelay
$3.60
Mistral AI 原价
$12.00
每月节省$8.40
每年 $100.80
概览
Ministral 3 14B 是什么?
70% below Mistral's list price.：Ministral 3 14B is the largest model in Mistral AI's Ministral 3 family, released on December 2, 2025: an open-weight model that reads text 和 images and, by Mistral's account, performs comparably to its larger Mistral Small 3.2 24B. The Zurelay Ministral 3 14B API serves its instruct version through one OpenAI-compatible endpoint with a 131,072-token context window,
Mistral AI released the Ministral 3 family on December 2, 2025, alongside Mistral Large 3, in three sizes: 3B, 8B 和 14B. Each size comes as a base, an instruct and a reasoning model, all with image understanding and all under the Apache 2.0 license, with the weights on Hugging Face. Ministral 3 14B pairs a 13.5B-parameter language model with a 0.4B-parameter vision encoder, and Mistral says it fits in 24GB of GPU memory at FP8. On Zurelay the model ID is ministral-3-14b, and it 每 1M 输出 token 的费用为 $0.06 per 1M input tokens and $0.06。
Mistral built Ministral 3 for edge 和 local use, 和 points the 14B model at private chat and assistant deployments on limited hardware, local agents and fine-tuning. It handles dozens of languages, including English, French, Spanish, German, Italian, Portuguese, Dutch, Chinese, Japanese, Korean and Arabic. Mistral says its Ministral instruct models often write an order of magnitude fewer tokens than comparable models.
On Mistral's model card the instruct model scores 0.551 on Arena Hard, 68.5 on WildBench 和 0.904 on MATH, ahead of Qwen3 14B without thinking (0.427, 65.1 和 0.870) and Gemma 3 12B (0.436, 63.2 and 0.854). On MM MTBench, a multimodal chat test, it scores 8.49 against 6.70 for Gemma 3 12B. These are Mistral's numbers, not independent ones.
The instruct model answers directly without reasoning first, so reasoning_effort has no effect. It calls tools, several at once 和 while streaming, 和 returns JSON for json_object or json_schema in response_format. Mistral lists a 256K context window on its own API; on Zurelay a request holds up to 131,072 tokens. In our tests it wrote about 170 tokens a second and knew little about events after 2024, so put the facts it needs in the prompt. You pay $0.06 per 1M input tokens and $0.06 per 1M output tokens against Mistral's list price of $0.20 and $0.20.
优势
Ministral 3 14B 的出色之处。 以及团队用它构建的应用。
图像理解
MM MTBench in Mistral's tests. Send screenshots, charts 和 photos next to your text. 上的 8.49
工具 和 JSON
Calls several tools at once, also while streaming, 和 follows a JSON schema, for agent steps 和 extraction.
开放权重
FP8, so what you build on Zurelay carries over to a self-hosted copy later.：Apache 2.0 weights on Hugging Face that fit in 24GB / GPU memory
使用场景
Assistants 和 chat
a very low price per token.：Customer-facing chat 和 internal assistants that need good everyday answers
Agent steps
Tool-calling steps such as looking things up, filling forms 和 choosing the next action, with JSON for the code that follows.
Extraction from text and 图像
Pull fields from emails, documents 和 screenshots into a JSON schema.
Multilingual work
Translate, summarize 和 answer in the languages Mistral lists, from French 和 German to Chinese, Japanese and Arabic.
开始使用
三步调用 Ministral 3 14B。
无需排队，也无需换用新的 SDK。只要你的代码能调用 OpenAI, 就能调用 Zurelay。
- 1
创建密钥
Sign up, add credit 和 create an API key. Set a monthly budget or a rate limit per key if you like.
- 2
Zurelay：Point your SDK
Change the base URL. Everything else in your code stays the same.
https://api.zurelay.com/v1 - 3
Use ministral-3-14b
Send chat completions as usual. Streaming, tool calls 和 usage reporting work as you expect.
ministral-3-14b
对比
相似模型。 同一个 key 即可调用。
70%优惠
Ministral 3 3B
每 1M token $0.03 / $0.03
half the price per token, about twice as fast in our tests.：Pick Ministral 3 3B for the simplest, highest-volume steps
Ministral 3 3B API53%优惠
Qwen 3.8 Flash
每 1M token $0.07 / $0.22
Pick Qwen 3.8 Flash for a 1M-token context, video input 和 a thinking mode when a task needs more than a small model.
Qwen 3.8 Flash API61%优惠
Gemini 3.1 Flash-Lite
每 1M token $0.10 / $0.59
Pick Gemini 3.1 Flash-Lite for Google's low-cost model with a 1M-token context 和 audio, video and PDF input.
Gemini 3.1 Flash-Lite APIMinistral 3 14B API 如何收费?
On Zurelay, ministral-3-14b costs $0.06 per 1M input tokens 和 $0.06 per 1M output tokens. Mistral's list price is $0.20 input 和 $0.20 output per 1M, so you save 70%. One rate applies at every prompt length, and failed requests are never billed.
如何通过 OpenAI SDK 调用 the Ministral 3 14B API?
Install the official openai package, set base_url to https://api.zurelay.com/v1 和 use your Zurelay API key. Then send a Chat Completions request with model set to "ministral-3-14b". Zurelay also accepts ministral-3-14b-instruct, mistralai/ministral-3-14b 和 ministral-14b. Mistral recommends a temperature of 0.1 or lower for production work.
Ministral 3 14B context window 是什么?
On Zurelay a request holds up to 131,072 tokens. Mistral lists a 256K window on its own API. The model reads text 和 images 和 writes text.
Can Ministral 3 14B read 张图片s or PDFs?
It reads 张图片s: send them in image_url parts, as https links or base64 data URLs, 和 they count as input tokens. Mistral suggests images close to square, with very wide or thin ones cropped. Zurelay doesn't take PDFs, audio or video for Ministral 3 14B: a request that includes one is turned away with a clear error before anything is billed.
Does Ministral 3 14B reason before it answers?
No. Zurelay serves the instruct model, which answers directly, so reasoning_effort has no effect 和 there are no reasoning tokens 至 pay for. Mistral also publishes a separate Ministral 3 14B Reasoning model, which Zurelay doesn't serve. For answers with thinking, use Qwen 3.8 Flash or DeepSeek V4.1 Flash.
Does Ministral 3 14B support tool calling 和 JSON output?
Yes. Pass tools as you would to any OpenAI model; it can call several at once, also while streaming. A tool forced by name is called, 和 tool_choice "none" works. Tool call ids must be at least 9 characters long, as the model's own 和 OpenAI's are. For JSON, send response_format with json_object or json_schema and the answer comes back as JSON.
Should I use Ministral 3 14B or Ministral 3 3B?
Both read text 和 images, hold 131,072 tokens on Zurelay 和 call tools. 14B is the stronger model, with a 13.5B-parameter language model against 3.4B, and scores higher in Mistral's tests: 0.551 against 0.305 on Arena Hard and 68.5 against 56.8 on WildBench. 3B costs half as much and was about twice as fast in our tests, so it suits the simplest, highest-volume steps.
Is Ministral 3 14B open source?
Its weights are open under the Apache 2.0 license, which allows commercial use 和 changes. Mistral publishes base, instruct 和 reasoning versions on Hugging Face and says the 14B fits in 24GB of GPU memory at FP8. The API is the quick way to use it without running your own hardware.
Is this the real Ministral 3 14B?
Yes. Requests for ministral-3-14b run on Mistral AI's Ministral 3 14B Instruct, 和 Zurelay doesn't swap in a different model. Only smart routing, which you turn on yourself, can answer with a close alternative when the model is down, 和 the x-zurelay-model header names the model that answered. Responses are sampled, so wording varies between runs. Zurelay is independent and is not affiliated with or endorsed by Mistral AI.
Zurelay 上的更多模型
全部模型- Zurelay Auto API每 1M token $0.025 / $0.12 起自动
- Claude Opus 5.5 API每 1M token $1.40 / $7.00−65%
- GPT-6 Astra API每 1M token $2.50 / $12.50−75%
- Nano Banana 2.1 API每张图片 $0.03 起−11%
- Nano Banana Pro API每张图片 $0.035 起−74%
- Nano Banana 2 API每张图片 $0.025 起−63%
- Nano Banana 2 Lite API$0.016 per 图片−52%
- Claude Sonnet 5.5 API每 1M token $0.79 / $3.95−61%
- Claude Haiku 5.5 API每 1M token $0.04 / $0.20−60%
- GPT-6.1 Sol API每 1M token $0.50 / $2.50−75%
- GPT-6.1 Terra API每 1M token $0.25 / $1.25−75%
- GPT-6 Sol API每 1M token $0.50 / $2.50−75%
- GPT-6 Luna API每 1M token $0.025 / $0.125−75%
- Grok 4.7 API每 1M token $0.70 / $2.10−65%
- Muse Wonder API每 1M token $1.20 / $6.00−70%
- MiMo V2.6 Pro API每 1M token $0.32 / $0.64−26%
- MiMo V2.6 Flash API每 1M token $0.075 / $0.15−46%
- DeepSeek V4.1 Flash 1017 API每 1M token $0.03 / $0.12−90%
- DeepSeek V4.1 Flash API每 1M token $0.03 / $0.12−90%
- DeepSeek V4 Flash API每 1M token $0.03 / $0.12−90%
- DeepSeek V4 Flash 0731 API每 1M token $0.029 / $0.087−69%
- DeepSeek V4 Pro 0813 API每 1M token $0.23 / $0.62−84%
- Claude Fable 5.1 API每 1M token $3.50 / $17.50−65%
- Claude Fable 5 API每 1M token $3.50 / $17.50−65%
- Claude Opus 5 API每 1M token $1.75 / $8.75−65%
- Claude Opus 4.8 API每 1M token $2.40 / $12.00−52%
- Claude Opus 4.7 API每 1M token $1.78 / $8.93−64%
- Claude Opus 4.6 API每 1M token $2.40 / $12.00−52%
- Claude Haiku 4.5 API每 1M token $0.54 / $2.70−46%
- GPT-5.6 Sol API每 1M token $1.14 / $5.70−72%
- GPT-5.6 Terra API每 1M token $0.54 / $3.26−73%
- GPT-5.6 Luna API每 1M token $0.08 / $0.49−59%
- Gemini 4 Argon API每 1M token $0.80 / $4.00−60%
- Gemini 3.1 Pro API每 1M token $0.50 / $3.45−71%
- Gemini 3.8 Flash API每 1M token $0.22 / $1.10−71%
- Gemini 3.7 Flash API每 1M token $0.32 / $1.59−58%
- Gemini 3.5 Flash-Lite API每 1M token $0.12 / $1.04−58%
- Gemini 3.1 Flash-Lite API每 1M token $0.10 / $0.59−61%
- GLM-5.3 API每 1M token $0.38 / $1.21−73%
- GLM-5.2 API每 1M token $0.38 / $1.21−73%
- GLM-5.3 Flash API每 1M token $0.058 / $0.19−62%
- GLM-5.3 FlashX API每 1M token $0.12 / $0.40−68%
- Kimi K3.1 Pro API每 1M token $0.80 / $4.00−60%
- Kimi K3.1 Pro Highspeed API每 1M token $3.30 / $16.50−18%
- Kimi K3 API每 1M token $1.27 / $6.37−58%
- Qwen 3.8 Max API每 1M token $1.20 / $3.60−40%
- Qwen 3.8 Flash API每 1M token $0.07 / $0.22−53%
- Qwen 3.8 Omni Flash API每 1M token $0.075 / $0.235−50%
- MiniMax M3.1 Flash API每 1M token $0.06 / $0.24−60%
- HY4 Preview API每 1M token $0.40 / $1.20−52%
- MiniMax M3 API每 1M token $0.13 / $0.51−57%
- 第 5 Preview 步 API每 1M token $0.25 / $0.68−75%
- GPT Image 2.5 Flare API每张图片 $0.028 起−47%
- GPT Image 2.5 Sunburst API每张图片 $0.026 起−51%
- GPT Image 2 API$0.02 per 图片−62%
- Grok Imagine Image 2.0 API每张图片 $0.045 起−25%
- FLUX 3 Image API每张图片 $0.04 起−17%
- Seedance 2.5 API每秒 $0.135 起视频
- Seedance 2.0 API每秒 $0.097 起视频
- Seedance 2.0 Fast API每秒 $0.074 起视频
- Kling 4.0 API每秒 $0.15 起视频
- Seedance 2.0 Mini API每秒 $0.047 起视频
- Gemini Omni Flash API每秒 $0.042 起视频
- Kling 3.0 API每秒 $0.125 起视频
- Wan 3.0 API每秒 $0.039 起视频
- Grok Imagine Video 1.5 API每秒 $0.071 起视频
- Veo 3.1 Fast API每秒 $0.025 起视频
- Vidu Q3 Turbo API每秒 $0.03 起视频
- PixVerse V6 API每秒 $0.019 起视频
- MiniMax H3 API每秒 $0.039 起视频
- Gemini 3.8 Flash TTS API每 1M token $0.30 / $5.40−40%
- Gemini 3.1 Flash TTS API每 1M token $0.60 / $12.00−40%
- Gemini 2.5 Pro TTS API每 1M token $0.60 / $12.00−40%
- Gemini 2.5 Flash TTS API每 1M token $0.30 / $6.00−40%
- Jev 1.14 API$0.024 per 1M 输入 token−43%
- Jev 1.13 API$0.035 per 1M 输入 token−17%
- Kimi K2.6 API每 1M token $0.30 / $1.30−68%
- Kimi K2.5 API每 1M token $0.20 / $0.95−68%
- MiniMax M2.5 API每 1M token $0.10 / $0.39−68%
- Ministral 3 3B API每 1M token $0.03 / $0.03−70%
- Muse Spark 1.3 Contributor API每 1M token $0.04 / $0.08−60%
- Abliterated Large V2 API每 1M token $2.90 / $4.85−3%