MiniMax M2 Models by MiniMax

MiniMaxMITOpen source
MiniMax releases · 10 in the last 12 months · this family litChangelog →
13 models2024–2026Up to 240k ctxFrom $0.150/1M input

Last refreshed 2026-06-15. Next refresh: weekly.

Details

ResearcherMiniMax
LicenseMITOSI-approved
Commercial useCommercial use: permitted
Models13
Released2024–2026
Max context240k

Capabilities

Reasoning3 of 13 models
JSON / Tool use3 of 13 models
Structured Outputs6 of 13 models

About

MiniMax M2 is MiniMax’s open-weight MoE model built for coding and agent workflows. The company frames it as a compact, high-throughput design (about 230B total parameters with about 10B activated) for end-to-end developer loops, multi-step tool use across shell, browser, and code runners, and general agent tasks. Weights are published for local serving (for example via vLLM or SGLang), and MiniMax also offers M2 through its Agent product and API platform.

Decision facts

Best fit
reasoningJSON / Tool usestructured outputs
Capability starting point
MiniMax M2.7 Highspeed with 205k context and reasoning, JSON / Tool use, and structured outputs
Lowest tracked input
MiniMax M2.5 · $0.150/1M · OpenRouter
Closest related family
MiniMax Image

Current Variants

Use-when guidance is based on each model's tracked capabilities, context window, release date, and replacement status.

13 in view
MiniMax M2.7Current

Use when the workload needs 205k context, reasoning, and JSON / Tool use.

2026-03205k contextreasoningJSON / Tool use
MiniMax M2.7 HighspeedCurrent

Use when the workload needs 205k context, reasoning, and JSON / Tool use.

2026-03205k contextreasoningJSON / Tool use
MiniMax M2-herCurrent

Use when the workload needs 64k context.

2026-0364k context
Together MiniMax M2.5Current

Use when the workload needs 200k context and 228.7B parameters.

2026-02200k context228.7B parameters
MiniMax M2.5 HighspeedCurrent

Use when the workload needs 205k context, reasoning, and JSON / Tool use.

2026-02205k contextreasoningJSON / Tool use
MiniMax M2.1Current

Use when the workload needs 200k context and structured outputs.

2025-12200k contextstructured outputs
MiniMax-M2-80kCurrent

Use when the workload needs 80k context and 80B parameters.

2025-1080k context80B parameters
MiniMax M2Current

Use when the workload needs 197k context and structured outputs.

2025-10197k contextstructured outputs
MiniMax M2.5Current

Use when the workload needs 197k context and structured outputs.

2025-03197k contextstructured outputs
MiniMax-M2-240kCurrent

Use when the workload needs 240k context and 240B parameters.

2025-01240k context240B parameters
MiniMax-M2-240k-0905Current

Use when the workload needs 240k context and 240B parameters.

2025-01240k context240B parameters
MiniMax-M2-80k-0905Current

Use when the workload needs 80k context and 80B parameters.

2025-0180k context80B parameters
MiniMax-M2.5Current

Use when provider availability and model metadata match the workload.

2024-09

Release Timeline

7 release groups
2026-03
3 current
MiniMax M2-her
64k context
Current
MiniMax M2.7
205k contextreasoningJSON / Tool use
Current
MiniMax M2.7 Highspeed
205k contextreasoningJSON / Tool use
Current
2026-02
2 current
MiniMax M2.5 Highspeed
205k contextreasoningJSON / Tool use
Current
Together MiniMax M2.5
200k context228.7B parameters
Current
2025-12
1 current
MiniMax M2.1
200k contextstructured outputs
Current
2025-10
2 current
MiniMax M2
197k contextstructured outputs
Current
MiniMax-M2-80k
80k context80B parameters
Current
2025-03
1 current
MiniMax M2.5
197k contextstructured outputs
Current
2025-01
3 current
MiniMax-M2-240k
240k context240B parameters
Current
MiniMax-M2-240k-0905
240k context240B parameters
Current
MiniMax-M2-80k-0905
80k context80B parameters
Current
2024-09
1 current
MiniMax-M2.5
Current

Specifications(13 models)

MiniMax M2 model specifications comparison
ModelReleasedContextParametersReasoningJSON / Tool useStructured Outputs
MiniMax M2.72026-03205k10B activeYesYesYes
MiniMax M2.7 Highspeed2026-03205k10B activeYesYesYes
MiniMax M2-her2026-0364k—NoNoNo
Together MiniMax M2.52026-02200k228.7BNoNoNo
MiniMax M2.5 Highspeed2026-02205k230B (10B active)YesYesYes
MiniMax M2.12025-12200k—NoNoYes
MiniMax-M2-80k2025-1080k80BNoNoNo
MiniMax M22025-10197k230B (10B active)NoNoYes
MiniMax M2.52025-03197k230B (10B active)NoNoYes
MiniMax-M2-240k2025-01240k240BNoNoNo
MiniMax-M2-240k-09052025-01240k240BNoNoNo
MiniMax-M2-80k-09052025-0180k80BNoNoNo
MiniMax-M2.52024-09—230B (10B active)NoNoNo

Pricing

MiniMax M2 model pricing by provider
ModelProviderInput / 1MOutput / 1MType
MiniMax M2.5OpenRouter$0.15$1.15Serverless
MiniMax M2OpenRouter$0.255$1Serverless
MiniMax M2.7OpenRouter$0.279$1.2Serverless
MiniMax M2.1OpenRouter$0.29$0.95Serverless
MiniMax M2.5Featherless$0.295$1.2Serverless
MiniMax-M2-240kFireworks AI$0.3$1.2Serverless
MiniMax-M2-80k-0905Fireworks AI$0.3$1.2Serverless
MiniMax-M2-240k-0905Fireworks AI$0.3$1.2Serverless
MiniMax-M2.5Fireworks AI$0.3$1.2Serverless
MiniMax M2.5Together AI$0.3$1.2Serverless
MiniMax M2.1AWS Bedrock$0.3$1.2Serverless
MiniMax M2.7Fireworks AI$0.3$1.2Serverless
MiniMax M2Vercel AI Gateway$0.3$1.2Serverless
MiniMax M2.1Vercel AI Gateway$0.3$1.2Serverless
MiniMax M2.5Vercel AI Gateway$0.3$1.2Serverless
MiniMax M2.7Vercel AI Gateway$0.3$1.2Serverless
MiniMax M2.7Novita AI$0.3$1.2Serverless
MiniMax M2.5Novita AI$0.3$1.2Serverless
MiniMax M2.1Novita AI$0.3$1.2Serverless
MiniMax M2Novita AI$0.3$1.2Serverless
MiniMax M2-herOpenRouter$0.3$1.2Serverless
MiniMax M2AWS Bedrock$0.4$1.6Serverless
MiniMax M2.5 HighspeedVercel AI Gateway$0.6$2.4Serverless
MiniMax M2.7 HighspeedVercel AI Gateway$0.6$2.4Serverless
MiniMax M2.5 HighspeedNovita AI$0.6$2.4Serverless
MiniMax-M2-80kFireworks AI$0.9$0.9Serverless
MiniMax M2Fireworks AI$0.9$0.9Serverless

Popular comparisons in this family