impulse

Ministral 3B

Open weights
API id ministral-3b-latest
Azure AI Foundry retires it on 9 Dec 2026 (in 21 days). 2 hosts keep serving it.
Input / 1M
$0.10
at Mistral AI
Output / 1M
$0.10
at Mistral AI
Context window
131K
tokens
Max output
131K
tokens per reply
Hosts
3
3 listings
Released
-
date not published

Pricing

List prices at Mistral AI, in US dollars per million tokens.

TierInput / 1MOutput / 1M
Standard$0.10$0.10
Cached input
90% off input
$0.01

What it costs a month

At Mistral AI list prices, 30 days.

  • Chat assistant5,000 requests a day, 2K in / 0.5K out, cached$29.40
  • RAG over documents2,000 requests a day, 8K in / 0.6K out, cached$30.00
  • Coding agent1,000 requests a day, 40K in / 2K out, cached$50.40
  • Overnight batch20,000 requests a day, 4K in / 0.4K out$264
Use your own numbers

Where to run it

3 hosts serve Ministral 3B. The comparison is on blended price, 3 parts input to 1 part output.

HostIn / out per 1M
Mistral AILab API
ministral-3b-latest
$0.10 / $0.10
ministral-3b
$0.04 / $0.04
−60% vs headline
Retires 9 Dec 2026
ministral-3b
$0.04 / $0.04
−60% vs headline

Capabilities

Core

  • Streaming
  • Function calling
  • JSON mode
  • Structured outputs

Multimodal

  • Vision
  • Image input
  • Audio
  • Image generation
  • Video

Advanced

  • Reasoning
  • Tool calling
  • MCP
  • Prompt caching
  • Batch API
  • Embeddings
Takes in
TextImages
Produces
Text
Features
Assistant prefill
Open weights
License: Apache-2.0
Runtimes: Ollama, llama.cpp, vLLM

Cheaper alternatives

Current models that cost less and keep most of what this one can do: tools, vision, reasoning, structured output, caching and enough context.

Same model, cheaper host

  • Azure AI Foundry
    ministral-3b
    $0.04 / $0.04
    60% less
  • Vercel AI Gateway
    ministral-3b
    $0.04 / $0.04
    60% less

Smaller Mistral AI models

No cheaper Mistral AI model keeps its key capabilities.

Other frontier labs

No cheaper match from another frontier lab.

Open-weight picks

  • GLM 5.3 Flash
    Zhipu AI · on CLF AI Gateway
    $0.06 / $0.20
    Has every key capability
    5% less
  • DeepSeek V4.1 Flash Fast
    DeepSeek · on Refuseless
    $0.05 / $0.15
    Has every key capability
    25% less
  • Qwen3.7 Flash Variant1
    Alibaba Qwen · on EmpirioLabs AI
    $0.028 / $0.11
    Has every key capability
    52% less
  • Gemma3 4B
    Google · on LlamaGate
    $0.03 / $0.08
    Has every key capability
    58% less

Call it

Using Mistral AI's API.

Python
from litellm import completion

response = completion(
    model="mistral/ministral-3b-latest",
    messages=[{"role": "user", "content": "Hello"}],
)
print(response.choices[0].message.content)

Quick answers

How much does Ministral 3B cost?

On Mistral AI, Ministral 3B costs $0.10 per million input tokens and $0.10 per million output tokens. Cached input costs $0.01 per million tokens. The cheapest host is Azure AI Foundry, at $0.04 per million input tokens and $0.04 per million output tokens.

Which APIs offer Ministral 3B?

Ministral 3B runs on 3 APIs: Mistral AI, Azure AI Foundry and Vercel AI Gateway.

What is Ministral 3B's context window?

131K tokens, with up to 131K tokens in one reply.

Is Ministral 3B being retired?

Azure AI Foundry retires it on 9 Dec 2026. 2 hosts keep serving it.

What are cheaper alternatives to Ministral 3B?

Models that cost less and keep most of what it can do: GLM 5.3 Flash on CLF AI Gateway (5% less). Ministral 3B itself costs 60% less on Azure AI Foundry. Savings are on a mix of 3 input tokens to 1 output token.

What is Ministral 3B's API model id?

Use ministral-3b-latest with Mistral AI. Other hosts use their own ids: ministral-3b on Azure AI Foundry, ministral-3b on Vercel AI Gateway.

Sources

Raw catalog entry for ministral-3b-latest on Mistral AI
{
  "input_cost_per_token": 1e-7,
  "litellm_provider": "mistral",
  "max_input_tokens": 131072,
  "max_output_tokens": 131072,
  "mode": "chat",
  "output_cost_per_token": 1e-7,
  "source": "https://docs.mistral.ai/models/ministral-3-3b-25-12",
  "supports_assistant_prefill": true,
  "supports_function_calling": true,
  "supports_response_schema": true,
  "supports_tool_choice": true,
  "supports_vision": true,
  "cache_read_input_token_cost": 1e-8
}