MiniMax M3

MiniMaxKey minimax-m3Also reported as minimaxai/minimax-m3, minimax/minimax-m3, mnm/minimax-m3, dynamic/minimax-m3, minimax-cn-coding-plan/minimax-m3, min/minimax-m3, tr/minimax-m3, cmc/minimaxai/minimax-m3, custom/minimax-m3, cline-pass/minimax-m3, tokenrouter/minimax-m3, ollama/minimax-m3, fta/mmx/minimax-m3, 9router/minimax/minimax-m3, vy/minimax-m3, kimchi/minimax-m3, combo/minimax-m3, synthetic/hf:minimaxai/minimax-m3, fireworks/minimax-m3, MiniMax/MiniMax-M3

Rank: #62; by active users
Active users: 5; Down 2 from 7 active users in the previous 7 days. vs previous 7 days
Adoption: 0.8%; of 610 accounts
Tokens: 917.9M; cache included
Est. cost: $5; at public list prices
First seen: May 31, 2026; in any submission

Nov 6–12, 2026 (UTC)610 accounts100% of tokens from an identified modelData through Nov 12, 2026, 08:00 UTCHow we count

Daily active users

Distinct accounts that used this model on each UTC day.

1 active users on Nov 12, 2026

Daily active users for MiniMax M3, Nov 6, 2026 to Nov 12, 2026. Peak 3. Use the left and right arrow keys to read each day.320

Show daily values

Daily active users and tokens for MiniMax M3
Date (UTC)	Active users	Tokens
Nov 6, 2026	3	322.3M
Nov 7, 2026	3	471.6M
Nov 8, 2026	2	17.9M
Nov 9, 2026	2	18M
Nov 10, 2026	2	26.1M
Nov 11, 2026	2	50M
Nov 12, 2026	1	11.8M

Token mix

How the 917.9M tokens split. Cache reads are usually most of an agent's tokens and are priced far below input.

Input16%
Output0.3%
Cache read84%
Cache write0%
Reasoning<0.1%

Agents that carry MiniMax M3

Each agent's share of this model's tokens, and how many accounts used the model through it.

MiniMax Code80% of this model's tokens1active users
OpenClaw13% of this model's tokens1active users
Claude Code6.6% of this model's tokens2active users
Hermes Agent<0.1% of this model's tokens1active users
cc-mirror/python-urllib<0.1% of this model's tokens1active users
cc-mirror/curl<0.1% of this model's tokens1active users
cc-mirror/mozilla<0.1% of this model's tokens1active users

Benchmarks

The best result for MiniMax M3 in each benchmark Tokscale tracks. Each benchmark has its own unit, run conditions and source, so compare a score only with other models on the same benchmark. Rank is among the models Tokscale matched in that benchmark.

All benchmarks

Epoch Capabilities IndexOverall · Epoch AI
147.7
Epoch Capabilities Index: 147.7
Model
MiniMax M3
Listed as
MiniMax-M3
Conditions
Not reported
Interval
145.6–150.0
Unit
ECI
Model released
Jun 1, 2026
Retrieved
Nov 12, 2026
Source · Epoch AI, ‘Capabilities & benchmarking’, epoch.ai/benchmarks (CC BY 4.0)
Rank #76 of 158 matched modelsNo conditions reportedModel released Jun 1, 2026Source: epoch.ai (opens in a new tab)
WebDev ArenaWeb development · LMArena
1,482
WebDev Arena: 1,482
Model
MiniMax M3
Listed as
MiniMax-M3
Conditions
Not reported
Interval
1477–1488
Unit
Arena score
Published
Oct 19, 2026
Retrieved
Oct 21, 2026
Source · LMArena leaderboard dataset (lmarena-ai/leaderboard-dataset on Hugging Face), CC BY 4.0
Rank #56 of 110 matched modelsNo conditions reportedPublished Oct 19, 2026Source: huggingface.co (opens in a new tab)
Agent ArenaAgents · LMArena
−0.09
Agent Arena: −0.09
Model
MiniMax M3
Listed as
MiniMax-M3
Conditions
Arena Agent Mode
Interval
-0.11 to -0.08
Unit
net improvement
Published
Oct 19, 2026
Retrieved
Oct 21, 2026
Source · LMArena leaderboard dataset (lmarena-ai/leaderboard-dataset on Hugging Face), CC BY 4.0
Rank #46 of 48 matched modelsArena Agent ModePublished Oct 19, 2026Source: huggingface.co (opens in a new tab)
GPQA DiamondReasoning · Epoch AI
90.9%
GPQA Diamond: 90.9%
Model
MiniMax M3
Listed as
MiniMax-M3
Conditions
Not reported
Unit
% correct
Run date
Aug 10, 2026
Retrieved
Nov 12, 2026
Best of 2 reported results for this model.
Source · Epoch AI, ‘Capabilities & benchmarking’, epoch.ai/benchmarks (CC BY 4.0)
Rank #37 of 160 matched modelsNo conditions reportedRun date Aug 10, 2026Source: epoch.ai (opens in a new tab)
Near its ceiling: the top score is 96.5%, and Epoch AI estimates the reachable maximum at 97%. Differences between top models are small.
Show 1 more result with other run conditionsHide the other results
- Listed as MiniMax-M3_none
  81.3%
  GPQA Diamond: 81.3%
  Model
  MiniMax M3
  Listed as
  MiniMax-M3_none
  Conditions
  No reasoning
  Unit
  % correct
  Run date
  Aug 10, 2026
  Retrieved
  Nov 12, 2026
  Source · Epoch AI, ‘Capabilities & benchmarking’, epoch.ai/benchmarks (CC BY 4.0)
  No reasoningRun date Aug 10, 2026Source: epoch.ai (opens in a new tab)

Pricing

List prices from models.dev in US dollars per 1M tokens, retrieved Nov 12, 2026. MiniMax (minimax.io) is the lab's own API and the reference price for the estimates below.

Price at real agent usageEstimate at list prices

What 1M tokens cost at the reference list price when they split into input, output and cache the way Tokscale users' agents used them. It is not what anyone paid.

At this model's mix: $0.10; Est. per 1M tokens
84% cache reads
At the community mix: $0.07; Est. per 1M tokens
96% cache reads
3:1 input/output blend: $0.52; Per 1M tokens at list prices
No cache

Cache reads were 84% of MiniMax M3 tokens in the last 7 days and 96% of all tokens. The 3:1 blend, (3 × input + output) ÷ 4, is the usual comparison and ignores cache.

Assumptions

MiniMax (minimax.io) list prices per 1M tokens: input $0.30, output $1.20, cache read $0.06.
This model's mix is its own input, output, cache and reasoning tokens in the last 7 days.
The community mix is the tokens of every identified model in the last 7 days.
This is an estimate at list prices, not what anyone paid.
Reasoning tokens are priced as output.
Long-context tiers, batch pricing and discounts are not applied.
Token mixes come from public Tokscale submissions. Hidden accounts are excluded.
No cache-write price is listed, so cache writes are priced as input.

List prices by provider

MiniMax (minimax.io)ReferenceMiniMax-M3
$0.30 input per 1M tokens$1.20 output per 1M tokens$0.06 cache read per 1M tokensCache write not listed1M tokens of context
OpenRouterminimax/minimax-m3
$0.30 input per 1M tokens$1.20 output per 1M tokens$0.06 cache read per 1M tokensCache write not listed1.05M tokens of context
Vercel AI Gatewayminimax/minimax-m3
$0.30 input per 1M tokens$1.20 output per 1M tokens$0.06 cache read per 1M tokensCache write not listed512K tokens of context
Together AIMiniMaxAI/MiniMax-M3
$0.30 input per 1M tokens$1.20 output per 1M tokens$0.06 cache read per 1M tokensCache write not listed524K tokens of context

Show 2 more providersShow fewer providers

Fireworks AIaccounts/fireworks/models/minimax-m3
$0.30 input per 1M tokens$1.20 output per 1M tokens$0.06 cache read per 1M tokensCache write not listed512K tokens of context
Deep InfraMiniMaxAI/MiniMax-M3
$0.28 input per 1M tokens$1.10 output per 1M tokens$0.06 cache read per 1M tokensCache write not listed524K tokens of context

Specs

From models.dev (MIT), retrieved Nov 12, 2026, as the lab or the first listed provider reports them. A dash means models.dev does not list the value.

Context window: 1M tokens
Max output: 524K tokens
Input: Text, Image, Video
Output: Text
Knowledge cutoff: — not listed
Release date: Jun 1, 2026
Open weights: Yes
Reasoning: Yes
Tool calling: Yes

Top public profiles

Public profiles that used MiniMax M3 most in this period, by tokens. Hidden accounts are never listed.

1jhzhang09@jhzhang09790.2Mtokens$4est. cost3active days$4 est. cost · 3 active days
2Dorian@Denszh119.6Mtokens—est. cost7active daysno price · 7 active days
3Fullstop000@Fullstop0008Mtokens<$1est. cost1active days<$1 est. cost · 1 active days
4PROTIK MONDAL@MNDL-2742.5Ktokens<$1est. cost2active days<$1 est. cost · 2 active days
5如是@MiFaZhan12.5Ktokens<$1est. cost2active days<$1 est. cost · 2 active days

Other MiniMax models in the same period, by active users.

MiniMax M3.1 Flash1.8% adoption11active users
MiniMax M3.1 Flash Cloud0.2% adoption1active users
MiniMax M3.1 Flash Preview0.2% adoption1active users
MiniMax M3.1 Flash Free0.2% adoption1active users
MiniMax M2.50.2% adoption1active users
MiniMax M3 Max0.2% adoption1active users

All models

Sources and attribution

Each score keeps its own source, unit and run conditions. Tokscale does not combine benchmarks into one score, and a missing score means the model was not evaluated, not that it scored zero.

Epoch AIEpoch Capabilities Index, Epoch's own runs and the external results it collects.CC BY 4.0 · Retrieved Nov 12, 2026 · Data through Nov 9, 2026Epoch AI, ‘Capabilities & benchmarking’. Published online at epoch.ai. Retrieved from https://epoch.ai/benchmarks.
LMArenaWebDev Arena and Agent Arena leaderboards.CC BY 4.0 · Retrieved Oct 21, 2026 · Data through Oct 19, 2026LMArena leaderboard dataset (lmarena-ai/leaderboard-dataset), Hugging Face.
models.devSpecs and list prices per provider.MIT · Retrieved Nov 12, 2026models.dev, an open catalog of AI models (github.com/anomalyco/models.dev).
TokscaleActive users, tokens and token mix from public Tokscale submissions. Hidden accounts are excluded.How we count