That's NeuralModel catalog

mistralai

Mistral: Ministral 3 8B 2512

mistralai/ministral-8b-2512

Our take

AI-assisted analysis based on the sources below. Suitability is an editorial assessment; test it on your own workload.

Mistral: Ministral 3 8B 2512 is the FP8 instruct member of Mistral's Ministral 3 family, listed on OpenRouter as mistralai/ministral-8b-2512. The model card describes a compact text-and-image-to-text model with a 256k-token context window, multilingual support, system-prompt adherence, function calling, and JSON output for edge-oriented deployment. OpenRouter lists text and image inputs with text output. Vendor and catalog benchmark figures are reported in the sources, not independently verified here.[1][2]

Strengths

  • Vendor documentation reports vision input, multilingual coverage, strong system-prompt adherence, native function calling, and JSON output.[2]
  • The card describes an FP8 instruct checkpoint for edge use, reported to fit within 12GB of VRAM in FP8 and less when further quantized.[2]
  • OpenRouter lists a 262,144-token context window and support for tools and structured outputs; the card reports an Apache 2.0 license.[1][2]

Limitations

  • Evidence is limited to a catalog entry and vendor documentation; validate quality, latency, and cost on your workload.[1][2]
  • Available benchmark figures are reported by the sources, and the OpenRouter entries cover only selected design, coding, and agentic categories.[1][2]
  • OpenRouter does not list a reasoning capability, and no knowledge cutoff or safety evaluation is supplied for this exact deployment.[1][2]

Where to use it

  • Consider for local or embedded multilingual chat assistants where a small footprint matters, subject to hardware and latency testing.[2]
  • Consider for image or document description and other workflows requiring text-plus-image input.[1][2]
  • Consider for tool-using assistants that need JSON or structured outputs, after validating call reliability.[1][2]

Sources

  1. [1] OpenRouter catalog · Mistral: Ministral 3 8B 2512Sources checked: Nov 13, 2026
  2. [2] Model documentation · Mistral: Ministral 3 8B 2512Sources checked: Nov 13, 2026

About this model

Description below is the original text supplied by OpenRouter, shown in the source language.

A balanced model in the Ministral 3 family, Ministral 3 8B is a powerful, efficient tiny language model with vision capabilities.

Specifications

Model ID
mistralai/ministral-8b-2512
Creator
mistralai
Added to OpenRouter
02.12.2025
Context length
262,144
Modality
text+image->text
Input modalities
text, image
Output modalities
text
Token pricing
USD per 1M tokens: in $0.150 / out $0.150
Knowledge cutoff
—
Expiration date
—

Tasks and capabilities

A grade appears only where the source supplies a matching measurement for that task, the model is active and the cohort holds at least 10 comparable models. The capability column lists facts the source declares — supported formats, reasoning and tool parameters — which are not measured quality.

TaskGradeSource-declared capability
General qualityNo measurement—
CodingCC — below the B band—
AgentsCC — below the B bandparameters: structured_outputs, tool_choice, tools
WritingNo measurementoutput: text
ReasoningNo measurement—
Document analysisNo measurement—
VisionNo measurementinput: image
Image generationNo measurement—
AudioNo measurement—
VideoNo measurement—
EmbeddingsNo measurement—

Evaluations

SourceBenchmarkValueGradeData retrieved
Artificial AnalysisCoding index9.7CC — below the B band13.11.2026
Artificial AnalysisAgentic index0.6CC — below the B band13.11.2026
Design Arenamodels · 3dwin rate 46.2% · rank 1171,014EloCC — below the B band13.11.2026
Design Arenamodels · codecategorieswin rate 42.9% · rank 1221,044EloCC — below the B band13.11.2026
Design Arenamodels · gamedevwin rate 38.7% · rank 128981EloCC — below the B band13.11.2026
Design Arenamodels · websitewin rate 42.9% · rank 1221,058EloCC — below the B band13.11.2026
How grades work

Each letter grade compares a model only against other models scored by the same source and metric in the current snapshot. S means the top 10%, A the next 20%, B the next 30%, C the rest. Models without a published measurement are shown as not rated — a supported format alone never implies measured quality.

  • S — S — top 10% of the cohort
  • A — A — next 20% of the cohort
  • B — B — next 30% of the cohort
  • C — C — below the B band

Providers

Optional live provider statistics from OpenRouter. Missing values are unknown, not zero.

ProviderContext lengthPrice in /1MPrice out /1MLatency p50 (30m, s)Throughput p50 (30m, tok/s)Uptime (30m)
Mistral262,144$0.150$0.150——100%
Mistral262,144$0.165$0.165——100%

Provider data retrieved: 13.11.2026

No published articles identify this model yet.