MiniMax M3

MiniMaxOpen weightsMiniMax Community LicenseReleased Jun 1, 2026

On the arena.ai Agents leaderboard, MiniMax M3 has a Net Improvement of -10.42% (—, rank #51, 95% interval -11.61%–-9.24%). MiniMax M3 ranks #104 on the LMArena Text leaderboard (style control) with an Arena score of 1435.9 (±3.7, 72,963 votes). Artificial Analysis Intelligence Index: 18.9. API price: $0.30 input / $1.20 output per 1M tokens. Context: 1.05M. Nearest rivals: GPT-5.2 (1437.4) above and DeepSeek V4 Flash (1435.8) below.

Quick answer

Agents leaderboard
#51 of 37 · -10.42% success
Text (LMArena)
#104 · 1435.9
17NAS overall
#35 of 36
API price (in / out)
$0.30 / $1.20 per 1M tokens
Closest rival
GPT-5.2 · 1437.4
Data updated
Nov 10, 2026 · synced daily

Sources: arena.ai, LMArena, Artificial Analysis, OpenRouter

Key metrics

Agent Net Improvement-10.42%44,355 sessions · #51
Arena score1435.9±3.7LMArena #104 · 72,963 votes
Intelligence18.9Variant: default
Coding24.6Artificial Analysis
Writing (creative)1394.3±6.2LMArena creative writing #112 / 419
List price in / out$0.30 / $1.20USD per 1M tokens
Lowest route$0.30 / $1.20Cheapest OpenRouter route
Context1.05Mtokens
Output speed106tok/sAA measured median
TTFT / first answer0.7 / 20sReasoning model: includes thinking

Rank history

LMArena rank, last 26 days
#95#100#105Oct 16Nov 10

#96 on Oct 16, #104 now; best #96 in this period.

Models ranked nearby

Sorted by Arena score
#ModelArenaIntel.Output $tok/s
56Grok 4.71442.035.5$6.0058
57Qwen3.5 397B1441.5–$3.60–
58GPT-5.21437.4–$14.00–
59MiniMax M3This page1435.918.9$1.20106
60DeepSeek V4 Flash1435.8–$0.28–
61MiMo V2.51433.7–$0.27–
62Qwen3.8 27B1433.420.1$3.0038

Pricing

  • Official API list price: $0.30 input and $1.20 output per 1M tokens (USD).
  • At a typical 3:1 input-to-output ratio, the blended cost is about $0.53 per 1M tokens.
  • Among 79 models with an output price, it is the #15 cheapest.
  • The cheapest OpenRouter route charges $0.30 input / $1.20 output.

Sources & methodology

On the arena.ai Agents leaderboard, MiniMax M3 has a Net Improvement of -10.42% (—, rank #51, 95% interval -11.61%–-9.24%).

  • arena.ai Agents — agent Net Improvement, 95% interval, sessions and rank (via the public Hugging Face dataset, CC BY 4.0).
  • LMArena — Arena score (Text leaderboard with style control, the same view arena.ai shows by default), 95% confidence interval, votes and rank.
  • Artificial Analysis — Intelligence / Coding Index, output speed and latency (medians).
  • OpenRouter — Route prices, context length and listing date.

Arena scores use the Bradley-Terry model (not Elo); ± is the half-width of the 95% confidence interval. Latency figures are Artificial Analysis medians; for reasoning models, time to first token and to first answer include the thinking phase. Data syncs daily at 14:20 (UTC+8).

Limits: Arena scores come from anonymous users voting on chats, so they reflect conversational preference rather than agent or coding benchmarks; with few votes the interval is wide and ranks move. Prices and speeds change as providers update them.

FAQ

How does MiniMax M3 do on the Agents leaderboard?

On the arena.ai Agents leaderboard, MiniMax M3 has a Net Improvement of -10.42% (—, rank #51, 95% interval -11.61%–-9.24%).

What rank is MiniMax M3 right now?

MiniMax M3 is #59 on our leaderboard of 70 scored models (Arena score 1435.9); its raw LMArena rank is #104.

What does MiniMax M3 cost via API?

Input $0.30 / output $1.20 per 1M tokens (official API list price; the OpenRouter quote when no official price is available). At a typical 3:1 input-to-output ratio, the blended cost is about $0.53 per 1M tokens.

How fast is MiniMax M3?

Artificial Analysis measures about 106 output tokens per second, 0.7 s to the first token and 20 s to the first answer token; it is a reasoning model, so latency includes thinking time.

How close is MiniMax M3 to the models ranked next to it?

It is 1.5 points behind GPT-5.2 (1437.4) and 0.1 points ahead of DeepSeek V4 Flash (1435.8); gaps smaller than the confidence interval (±3.7) are not significant.

Is MiniMax M3 open source? Can it be self-hosted?

MiniMax M3 is an open-weights model (license: MiniMax Community License); you can download the weights and self-host.

What is the context length of MiniMax M3?

1.05M (1,048,576 tokens, as listed on OpenRouter).

MiniMax M3 vs GPT-5.2: which is better?

On the LMArena Text leaderboard MiniMax M3 scores 1435.9 and GPT-5.2 scores 1437.4 (a 1.5-point gap, within the confidence interval, so not significant). On price, MiniMax M3 costs $1.20 and GPT-5.2 $14.00 per 1M output tokens.

Is there a cheaper model close to MiniMax M3?

Yes: DeepSeek V4 Flash scores 1435.8 on the Text leaderboard (MiniMax M3: 1435.9) and costs $0.28 per 1M output tokens, about 77% less than MiniMax M3 ($1.20).

Where does the MiniMax M3 data come from and how often is it updated?

Agent success rates come from the arena.ai Agents leaderboard (CC BY 4.0), Text scores from LMArena, intelligence and speed from Artificial Analysis, and prices and context from official list prices and OpenRouter. 17NAS syncs daily at 06:20 UTC; this page was last updated Nov 10, 2026.

Related leaderboards