Context
1M
Max output
1M
Serving providers
11
Cheapest input
$0.23 /1M
On-demand listed rates, OpenRouter routed quotes and Vercel reseller quotes. Enterprise commitments (committed-use discounts, provisioned throughput, negotiated rates) can be materially lower.
Benchmarks
independent evaluations · source: Artificial Analysis and Hugging Face leaderboardsSpeed & latency
Output speed
107 t/s
tokens / second
Time to first token
1.21s
latency
Headline indices
Individual benchmarks · click a row for effort variants
| Benchmark | Best score | Effort | Cost / task | |||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Aa Analyst Agent | 0.1 | default | — | |||||||||||||||||
| AA-BriefcaseUndated | 31% | default | — | |||||||||||||||||
| Automation Bench | 0.2 | default | — | |||||||||||||||||
| CritPt | 4% | default | — | |||||||||||||||||
| EvalComputeProxyFrozen Aug 18, 2026 | 4190.3 | default | — | |||||||||||||||||
| GDPval v2Frozen Aug 18, 2026 | 1386.9 | default | — | |||||||||||||||||
| ||||||||||||||||||||
| GPQA Diamond | 93% | default | — | |||||||||||||||||
| Harvey Lab | 0 | default | — | |||||||||||||||||
| Humanity's Last Exam | 39% | default | — | |||||||||||||||||
| IFBench | 83% | default | — | |||||||||||||||||
| Long-Context Reasoning | 83% | default | — | |||||||||||||||||
| Mlcr OverallFrozen Sep 9, 2026 | 0.2 | default | — | |||||||||||||||||
| MMMU-Pro | 79% | default | — | |||||||||||||||||
| Omniscience | 1.4 | default | — | |||||||||||||||||
| Omniscience Accuracy | 0.2 | default | — | |||||||||||||||||
| Omniscience Non Hallucination | 0.8 | default | — | |||||||||||||||||
| SciCode | 47% | default | — | |||||||||||||||||
| SWE-bench Pro | 59% | default | — | |||||||||||||||||
| SWE-bench Verified | 81% | default | — | |||||||||||||||||
| Terminal Bench40 | 0.0 | default | — | |||||||||||||||||
| ||||||||||||||||||||
| TerminalBench Hard | 42% | default | — | |||||||||||||||||
| TerminalBench v2.1 | 65% | default | — | |||||||||||||||||
| ||||||||||||||||||||
| τ-bench Banking | 15% | default | — | |||||||||||||||||
| τ²-bench | 89% | default | — | |||||||||||||||||
Artificial Analysis last published scores for this benchmark on Sep 9, 2026. Scores here are frozen at that date, and models released since are not on it.
We cannot tell whether this benchmark is still being published. Treat these scores as undated.
Artificial Analysis last published scores for this benchmark on Aug 18, 2026. Scores here are frozen at that date, and models released since are not on it.
Pricing detail
cost beyond the standard rate · source: models.devLong context (>200K)
$0.45 / $1.80
input / output · per 1M
Pricing by serving provider
standard tier · per 1M tokens · click a provider to expandInput pricing runs from $0.23 to $0.30 per 1M tokens across 11 providers, so the dearest route costs 30% more than the cheapest for the same model.
| Serving provider | Input /1M | Output /1M | Endpoints | |||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| WandbCheapest | $0.23 | $0.96 | 1 | |||||||||||||
Standardwandb/MiniMaxAI/MiniMax-M3
| ||||||||||||||||
| DeepInfra | $0.28 | $1.10 | 1 | |||||||||||||
Standarddeepinfra/MiniMaxAI/MiniMax-M3
| ||||||||||||||||
| Aihubmix | $0.288 | $1.15 | 1 | |||||||||||||
Standardaihubmix/minimax-m3
| ||||||||||||||||
| Fireworks AI | $0.30 | $1.20 | 1 | |||||||||||||
Standardfireworks_ai/accounts/fireworks/models/minimax-m3
| ||||||||||||||||
| Minimax | $0.30 | $1.20 | 1 | |||||||||||||
Standardminimax/MiniMax-M3
| ||||||||||||||||
| Nebius | $0.30 | $1.20 | 1 | |||||||||||||
Standardnebius/MiniMaxAI/MiniMax-M3
| ||||||||||||||||
| Novita | $0.30 | $1.20 | 1 | |||||||||||||
Standardnovita/minimax/minimax-m3
| ||||||||||||||||
| OpenRouter | $0.30 | $1.20 | 1 | |||||||||||||
Standardminimax/minimax-m3
| ||||||||||||||||
| Fireworks AI is dearest at $0.30 / $1.20 | ||||||||||||||||
Price history
listed seller prices, input and output per 1M tokens Input Output
Input down 62% since first tracked
Cost calculator
estimate your monthly spend on this model$123
estimated / month
Model IDs
copy the exact identifier for your platformminimax/minimax-m3:freewandb/MiniMaxAI/MiniMax-M3deepinfra/MiniMaxAI/MiniMax-M3aihubmix/minimax-m3fireworks_ai/accounts/fireworks/models/minimax-m3minimax/MiniMax-M3minimax/minimax-m3tencent/minimax-m3novita/minimax/minimax-m3together_ai/MiniMaxAI/MiniMax-M3minimax/minimax-m3:batchnebius/MiniMaxAI/MiniMax-M3minimax/minimax-m3
Frequently asked questions
MiniMax M3 pricing, context and availabilityHow much does MiniMax M3 cost?
MiniMax M3 costs $0.23 per 1M input tokens from Wandb, and $0.96 per 1M output tokens from Wandb. Across 11 serving providers, input prices range from $0.23 to $0.30 per 1M tokens.
What is the context window of MiniMax M3?
MiniMax M3 accepts up to 1M tokens of context and can return up to 1M output tokens.
Which providers serve MiniMax M3?
MiniMax M3 is available from 11 serving providers, each with its own pricing and model ID. Wandb is currently the cheapest.
What can MiniMax M3 do?
MiniMax M3 supports image input (vision), tool use, reasoning, prompt caching and structured output.
Other MiniMax models
compare pricing across the MiniMax lineupMiniMax M2.5$0.27/1M · 10 providersMiniMax M2.7$0.25/1M · 9 providersMiniMax M2.1$0.30/1M · 7 providersMiniMax M2$0.255/1M · 6 providersMiniMax M1 80k$0.10/1M · 2 providersMiniMax M2.1 Lightning$0.30/1M · 2 providersMiniMax M2.5 Highspeed$0.60/1M · 2 providersMiniMax M2.7 Highspeed$0.60/1M · 2 providers
Popular comparisons
head-to-head pages featuring MiniMax M3
MiniMax M3 vs Claude Fable 5side-by-side pricing & benchmarks MiniMax M3 vs Claude Fable 5.1side-by-side pricing & benchmarks MiniMax M3 vs Claude Opus 4.8side-by-side pricing & benchmarks MiniMax M3 vs Claude Opus 5side-by-side pricing & benchmarks MiniMax M3 vs Claude Sonnet 5side-by-side pricing & benchmarks Listed prices from LiteLLM, OpenRouter and Vercel · last checked Oct 14, 2026
Every weekday
AI moves fast. Here's your debrief.
News, analysis, tools, and more.
For people who build with AI