modelbenchmark.io

Ministral 3B (latest)

Mistral · ministral-3b

Compare

Compact Mistral model for edge, latency-sensitive, and cost-efficient workloads

Specification

mistral listing

Context
128K
Max output
128K
Released
2024-10-01
Knowledge cutoff
2024-03
Retires
Open weights
yes
Input
text
Output
text

Price

US dollars per million tokens · mistral list price

Input
$0.04
Output
$0.04
Cache read
$0.01
Cache write
$0.10

Quality

2 benchmarks · 1 source

Composite
4th
percentile of 334 scored models
Rank
320 / 334
±0.06 sd
Evidence
2 × 1
too little evidence to place with confidence
Effort range
one configuration only
reasoning0th1/4 bench
math9th1/7 bench

Every score

one row per source and configuration — nothing averaged away

BenchmarkScoreConfigurationSourceRun
MATH level 514.4 ±0.7defaultEpoch AI2025-01-27
GPQA diamond25.2 ±1.9defaultEpoch AI2025-01-27

The leading figure for a benchmark is the median across its configurations, so one heroic high-effort run cannot set the number.

Available from 13 hosts

HostIn $/MOut $/MCache rdCache wrContextOutputRetires
mistralvendor
$0.04
$0.10
$0.04
$0.10
$0.01
128K
131K
128K
131K
azure$0.04$0.04128K8K
azure_ai$0.04$0.04128K4K
azure-cognitive-services$0.04$0.04128K8K
vercel$0.04$0.04128K128K
vercel_ai_gateway · vercel-ai-gateway$0.04$0.04128K4K
kilo$0.10$0.10$0.01131K105K
llmgateway$0.10$0.10131K8K
llmgateway-providers$0.10$0.10131K131K
nano-gpt$0.10$0.10$0.05131K33K
openrouter$0.10$0.10$0.01131K
105K
131K
pioneer$0.10$0.10$0.10$0.10128K4K
cortecs$0.123$0.123$0.012256K256K

Price history

append-only observations · a listing writes a row only when its price moves

DateHostField$/M
2026-09-02cortecsCache read$0.011
2026-09-15cortecsCache read$0.012
2026-09-02cortecsInput$0.111
2026-09-15cortecsInput$0.123
2026-09-02cortecsOutput$0.111
2026-09-15cortecsOutput$0.123

More from Mistral

most-hosted first