LLM Pricing
ModelsLeaderboardsToolsProvidersReleasesArticles

© 2026 LLM Pricing

About
·Contact
·Privacy
·RSS
← Model list

MiniMax-M3

minimax·minimax/minimax-m3·BETA·Open weights·minimax series
MiniMax multimodal model for long-context coding, perception, and agent planning

Specs & pricing

Input / output per 1M tokens
Official price·MiniMax (minimax.io)
$0.30 / $1.20
Blended $0.53 · Cache read $0.06
Lowest paid·VultrCloud
$0.20 / $0.90
Blended $0.38 · 6× spread
1 more $0 channels
Context
1,000,000
Output limit
524,288
Knowledge cutoff
2025-01
Released / updated
2026-06-01 / 2026-06-01
Capabilities
✓ Reasoning✓ Tool use✓ Structured output✓ Temperature✓ Attachments
⚠ Providers report capability flags inconsistently
Modalities
TextImageVideoPDF
Weights
Hugging Face

MiniMax-M3 is offered by 70 providers on this page. Public prices are shown for 64 of them. Its official list price is $0.30 per 1M input tokens and $1.20 per 1M output tokens. The lowest paid channel is Vultr at $0.20 / $0.90 per 1M, about 1.4× below the list price. It also has 1 free ($0) channel and 4 covered by a paid subscription; free tiers usually carry rate limits, and subscription-covered access bills $0 per token only after the subscription fee.

Artificial Analysis rates it 18.9 on the Intelligence Index, with 27.9 for Strategy & Ops, its strongest domain. Against its lowest blended price of $0.375 per 1M, that is roughly 50 index points per dollar, which is the value ratio the leaderboards rank on. Median output speed is 89 tokens per second, with 0.78s to the first token. Running one task of the Artificial Analysis suite costs about $1.2851, which reflects how many tokens its reasoning consumes rather than the unit price alone.

The context window is 1,000,000 tokens at the reference host, but hosts report different limits, from 131,072 to 1,048,576, so the usable window depends on the provider you pick. It supports reasoning, tool use, and structured output. Accepted input modalities are Text, Image, Video, and PDF. The weights are open, so it can also be self-hosted or served through a gateway of your choice. Providers report the capability flags inconsistently, so verify a specific feature against the host you plan to use. Its training knowledge cuts off at 2025-01.

Quality & performanceArtificial Analysis · Intelligence Index v5

Intelligence18.9
Value50
Engineering26.2
Finance & Accounting25.1
Legal23.5
Healthcare & Medical21.7
Strategy & Ops27.9
Economics27.4
Output speed89 tok/s
TTFT0.78 s
Cost per task$1.29
Value formulaIQ 18.9 ÷ min blended $0.375 = 50

Quality is independently evaluated by Artificial Analysis. Speed/latency are model-level medians.

Available at 70 providers64 with public prices · 1 free · 4 subscription-covered

  • ProviderInput / Output
  • SCNet Token PlanGateway
    MiniMax-M3
    Subscription
    Cache read $0Context 1,048,576 ⚠Output limit 512,000
  • OpenCodeGateway
    minimax-m3-free
    Free
    Cache read $0Context 200,000 ⚠Output limit 32,000deprecated
  • Volcengine Ark Coding PlanGateway
    Subscription
    Cache read $0Context 1,048,576 ⚠Output limit 512,000deprecated
  • MiniMax Token Plan (minimax.cn)Gateway
    MiniMax-M3
    Subscription
    Cache read $0Context 1,000,000Output limit 524,288
  • MiniMax Token Plan (minimax.io)Gateway
    MiniMax-M3
    Subscription
    Cache read $0Context 1,000,000Output limit 524,288
  • VultrCloud
    $0.20 / $0.90
    Context 524,288 ⚠Output limit 512,000
  • EmpirioLabs AIGateway
    $0.23 / $0.90
    tiered >512K: $0.45
    Cache read $0.045Context 1,048,576 ⚠Output limit 524,288
  • CoreWeaveCloud
    MiniMaxAI/MiniMax-M3
    $0.23 / $0.96
    Cache read $0.05Context 262,144 ⚠Output limit 262,144
  • AmbientGateway
    $0.23 / $0.96
    Cache read $0.05Context 131,072 ⚠Output limit 131,072
  • VancineGateway
    MiniMax-M3
    $0.24 / $0.96
    Cache read $0.048Context 1,048,576 ⚠Output limit 512,000deprecated
  • MiniMax (minimax.io)OfficialFirst-party
    MiniMax-M3
    $0.30 / $1.20
    tiered >512K: $0.60
    Cache read $0.06Context 1,000,000Output limit 524,288
ProviderTierInputOutputCache readCache writeContextOutput limitStatus
SCNet Token Plan
MiniMax-M3
GatewaySubscription$0—1,048,576 ⚠512,000
OpenCode
minimax-m3-free
GatewayFree$0—200,000 ⚠32,000deprecated
Volcengine Ark Coding PlanGatewaySubscription$0—1,048,576 ⚠512,000deprecated
MiniMax Token Plan (minimax.cn)
MiniMax-M3
GatewaySubscription$0$01,000,000524,288
MiniMax Token Plan (minimax.io)
MiniMax-M3
GatewaySubscription$0$01,000,000524,288
VultrCloud$0.20$0.90——524,288 ⚠512,000
EmpirioLabs AIGateway$0.23
tiered >512K: $0.45
$0.90$0.045—1,048,576 ⚠524,288
CoreWeave
MiniMaxAI/MiniMax-M3
Cloud$0.23$0.96$0.05—262,144 ⚠262,144
AmbientGateway$0.23$0.96$0.05$0131,072 ⚠131,072
Vancine
MiniMax-M3
Gateway$0.24$0.96$0.048—1,048,576 ⚠512,000deprecated
MiniMax (minimax.io)Official
MiniMax-M3
First-party$0.30
tiered >512K: $0.60
$1.20$0.06—1,000,000524,288

Sorted by blended price (input×0.75 + output×0.25) asc. The official channel always shows regardless of rank. Whether a gateway's low price is actually usable can't be verified.

Reasoning control

Toggle (on / off)effort = loweffort = mediumeffort = higheffort = noneeffort = minimaleffort = xhigheffort = maxbudget_tokens

Interleaved thinking (reasoning between tool calls) is declared by 26 of 70 providers.

17 / 70 providers expose no reasoning control (reasoning_options: []).

Experimental modes2 items

ModeProviderInputOutputCache readCache write
priorityAbliteratedLLM$0.41$1.62$0.081—
priorityFireworks AI$0.45$1.80$0.09—

Your usage cost

Common scenarios
1EmpirioLabs AI$68.40
2CoreWeave$72.40
3Ambient$72.40
4Vancine$72.96
5AbliteratedLLM$82.08
6Boundless$84.12
12MiniMax (minimax.io) · Official$91.20

5 more channels offer $0 (SCNet Token Plan, OpenCode, Volcengine Ark Coding Plan etc.); free tiers usually have rate limits and no SLA, excluded from ranking.

Switch to EmpirioLabs AI to save $22.80/mo (25%)
Note: this is a gateway; verify availability and rate limits yourself.

Price historyone sample accumulated per data sync

Input listOutput listMin blended
$1.20$02026-08-212026-10-01
2026-10-01Input list $0.30Output list $1.20Min blended $0.38

Benchmark6 items

NameConditionsScoreMetricSource
SWE-Bench Verifiedharness: Claude Code80.5resolvedSource ↗
SWE-Bench Proharness: Claude Code59resolve rateSource ↗
Terminal-Benchv2.166success rateSource ↗
BrowseComp—83.52accuracySource ↗
MCP Atlas—74.2success rateSource ↗
OSWorld-Verified—70.06success rateSource ↗

The same benchmark scores very differently across harness / dataset, so the qualifying conditions must be shown together.

Artificial Analysis evaluations30 items

GPQA Diamond92.9%
Humanity's Last Exam39.0%
HLE-Diamond16.0%
MMMU-Pro78.5%
CritPt3.7%
CritPt v2 Hard28.0%
SciCode47.1%
FrontierCode23.0%
FrontierCode 2.021.9%
Terminal-Bench Hard42.4%
Terminal-Bench 2.165.2%
Terminal-Bench 4.02.0%
Terminal-Bench-Science0.5%
τ²-Bench Telecom88.9%
τ³-Bench Banking15.3%
AutomationBench-AA21.3%
AA-AnalystAgent10.0%
EnterpriseOps-Gym-AA32.1%
AA-Briefcase1089
GDPval-AA1250
GDP.pdf9.8%
AA-LCR83.0%
MLCR-AA4.4%
IFBench82.9%
AA-Omniscience Index-5.5
AA-Omniscience Accuracy44.9%
AA-Omniscience Non-Hallucination8.6%
Harvey LAB-AA0.0%
AA-3D746
Code2Video-AA724

Individual evaluations run by Artificial Analysis, on the same reasoning tier as the intelligence score above. Each benchmark has its own task set and scale (percent correct, points or an index), so rows are not comparable with one another.

Related models

MiniMax-M3.1-Flashsame series$0.15 / $0.60MiniMax H3same series—MiniMax H3 Maxsame series—GLM-5.3-Flashcheaper alternative$0.15 / $0.50DeepSeek V4.1 Flash 1017cheaper alternative$0.15 / $0.60Claude Haiku 5.5cheaper alternative$0.10 / $0.50GPT-6 Lunacheaper alternative$0.10 / $0.50MiniMax-M2.7-highspeedsame lab$0.60 / $2.40
Data partly from models.dev (MIT) · MiniMax (minimax.io) official docs ↗