Can I run MiniMax M2.1 locally?

Yes — the weights are open on Hugging Face under a Modified-MIT licence, so local deployment is permitted, though the full model is large and you should verify your hardware can handle it before downloading.

What does MiniMax M2.1 cost and what is its context window?

Via the MiniMax API it costs $0.30 per million input tokens and $1.20 per million output tokens, with a 204,800-token context window.

How does M2.1 differ from its predecessor M2?

M2.1 scores notably higher on software engineering benchmarks — 49.4% vs 36.2% on Multi-SWE-Bench and 72.5% vs 56.5% on SWE-Bench Multilingual — and MiniMax says outputs are cleaner and perceived response times are faster.

MiniMax M2.1

3782°

#llm #minimax #reasoning #coding #agents #multilingual

Run this model

Offers / Ads

OpenRouter Terminal / API

Pay per token

Chat with this model in the browser or call it from your code with one API key.

MiniMax | 204,800 token context | $0.30/M input, $1.20/M output. MiniMax-M2.1 is a lightweight, state-of-the-art large language model optimized for coding, agentic workflows, and modern application development. With only 10 billion activated parameters, it delivers a major jump in real-world capability while maintaining exceptional latency, scalability, and cost efficiency. Compared to its predecessor, M2.1 delivers cleaner, more concise outputs and faster perceived response times. It shows leading multilingual coding performance across major systems and application languages, achieving 49.4%…MiniMax | 204,800 token context | $0.30/M input, $1.20/M output. MiniMax-M2.1 is a lightweight, state-of-the-art large language model optimized for coding, agentic workflows, and modern application development. With only 10 billion activated parameters, it delivers a major jump in real-world capability while maintaining exceptional latency, scalability, and cost efficiency. Compared to its predecessor, M2.1 delivers cleaner, more concise outputs and faster perceived response times. It shows leading multilingual coding performance across major systems and application languages, achieving 49.4% on Multi-SWE-Bench and 72.5% on SWE-Bench Multilingual, and serves as a versatile agent “brain” for IDEs, coding tools, and general-purpose assistance. To avoid degrading this model's performance, MiniMax highly recommends preserving reasoning between turns. Learn more about using reasoning_details to pass back reasoning in our docsOpens in new tab.

Guides for this model

Best local LLM by use case (October 2026): the right open model for every job

How much VRAM do I actually need for local AI?

Open weights vs open source vs API-only: what you are actually allowed to do

What is a GGUF file? Quantized models explained (Q4, Q8, fp8, NF4)

All guides

Captain's notes

Written by Captain, the house editor

Notes from 2026-10-13

MiniMax M2.1 is a reasoning-capable language model from MiniMax with 10 billion activated parameters, a 200K-token context window, and open weights under a Modified-MIT licence. Its standout angle is multilingual software engineering: it outperforms Claude Sonnet 4.5 on Multi-SWE-Bench and SWE-Bench Multilingual while costing $0.30 per million input tokens. If you are building coding agents or IDE tooling and want something you can also run locally, it is worth a serious look.

Known for

Strong multilingual coding benchmark results, beating Claude Sonnet 4.5 on Multi-SWE-Bench.
Only 10B activated parameters, keeping latency low and self-hosting feasible.
Designed as an agentic 'brain': tool use, instruction following, long-horizon planning.
Open weights on Hugging Face under a Modified-MIT licence.
200K-token context at $0.30/$1.20 per million input/output tokens.

Watch out for

MiniMax strongly recommends preserving reasoning between turns; skipping this will degrade performance.
Modified-MIT is not standard MIT — check the licence terms before any commercial deployment.
Full model weights are large; confirm your VRAM before planning local use.

Best for

Prompts to try

Fix a real GitHub issue

Here is a Python repository and an open bug report. The bug: calling `process_batch()` with an empty list raises an unhandled `IndexError` instead of returning an empty result. Reproduce the error, identify the root cause in the code below, write a fix, and add a test case that would have caught it. [paste your code]

Build a multi-step agent

You have access to a web search tool and a code execution tool. A user wants to know the current exchange rate between USD and JPY, then calculate how much 3,750 USD is worth in JPY, and finally write the result to a local file called result.txt. Plan the steps, call the tools in order, and confirm the file was written.

Multilingual code review

Review the following TypeScript function for correctness, edge cases, and performance. Then rewrite it in idiomatic Rust, preserving the same behaviour. Explain any differences in error handling between the two versions. [paste your TypeScript function]

Compare with

MiniMax M1
API-only model
1,000,000 token context · $0.55/M input, $2.20/M output
MiniMax M3
Open-weights model
1,048,576 token context · $0.30/M input, $1.20/M output
MiniMax M2.5
Open-weights model
204,800 token context · $0.27/M input, $0.95/M output
MiniMax M2
Open-weights model
204,800 token context · $0.26/M input, $1.02/M output

More text models from minimax

Questions people ask

Can I run MiniMax M2.1 locally?: Yes — the weights are open on Hugging Face under a Modified-MIT licence, so local deployment is permitted, though the full model is large and you should verify your hardware can handle it before downloading.
What does MiniMax M2.1 cost and what is its context window?: Via the MiniMax API it costs $0.30 per million input tokens and $1.20 per million output tokens, with a 204,800-token context window.
How does M2.1 differ from its predecessor M2?: M2.1 scores notably higher on software engineering benchmarks — 49.4% vs 36.2% on Multi-SWE-Bench and 72.5% vs 56.5% on SWE-Bench Multilingual — and MiniMax says outputs are cleaner and perceived response times are faster.

Captain's verdict
A rare open-weights coding agent that punches above its activated-parameter count and undercuts closed rivals on price.

Model card on Hugging Face

Captain writes these notes from the model's own description and model card. Prices and context sizes are the provider's at the time of writing and can change.

Ranking badge

Updated Oct 22, 2026

Made this model? Show this badge on your page.

Paste it into your Hugging Face card or Civitai description.

Change colors / personalize / get code

Not this creator?

Get your own badge

How this score adds up

This model has no popularity data from any source yet, so it carries a stand-in number below 7,000° that keeps the lists complete. It never ranks above a model with data. The breakdown appears here as soon as a source reports it.

Glossary: the terms used here

Heat score: One number for how much a model is used and wanted right now, from 0° to 9,999°. Every list on this site that says hottest is sorted by it and nothing else.
Floor: The 7,000° every model with real data starts from. The scale runs from 0° to 9,999°; the part below the floor is reserved for stand-ins (models without data), so that a model with data always ranks above one without. The floor is the same for everyone and decides nothing; the degrees above it are the measurement.
Download-equivalents: Everything the score treats as a download, added up: downloads themselves, each torrent seeder as 1,000, each AI Horde image and each run on a generation site as one.
Log scale: Counting by tens, not by ones: going from 1,000 to 10,000 downloads adds the same degrees as going from 10,000 to 100,000. It keeps giants from crowding out everything else.
Likes: Thumbs up and likes from every source, worth half a download each on the log scale.
Adoption rate: How many downloads a day a model gets right now, from the difference between two readings of a source. A quarter of the weight of the download term.
Measured and estimated: Measured means two readings of a source exist and the difference is real. Estimated means only today's total is known and the past is spread over the model's life; the site says which one it is showing.
Seeder: Someone who keeps a copy of the weights online for others to download over a torrent, or a hosted mirror doing the same. Each one counts as 1,000 downloads.
AI Horde: A volunteer network that generates images for free on donated computers. The number of images it has made with a model is public, and each counts as one download.
Generation sites: Sites that host their own copies of a model and let people generate with it, such as PixAI, SeaArt and Yodayo. Their downloads count like Civitai's and each generation run as one download.
OpenRouter tokens: A text model served through an API, such as Claude or GPT, has no weights to download, so its usage is measured in tokens: the pieces of text it read and wrote through OpenRouter, a gateway many apps use, over the last 30 days. Every 10,000,000 tokens count as one download. Source: OpenRouter (openrouter.ai/rankings), CC BY 4.0.
Full model (checkpoint): The weights themselves, the file that generates on its own. It gets a bonus because add-ons depend on it.
LoRA and embedding: Small add-on files that steer someone else's weights toward a style, a character or a concept. They cannot run alone.
Family and weights: The base a model is built on, such as SD 1.5, SDXL, Flux or Wan. Every model inherits the release date of its family's weights.
Age: Degrees taken off for how long ago the family's weights came out: 150° a year, up to 750°. It keeps old downloads that had years to pile up from freezing the top of the list.
Family boost: A small, bounded nudge for LoRAs of a family this site is pushing forward right now, always listed in the breakdown.
Stand-in score: A stable number below the floor given to a model no source reports on yet, so lists stay complete. It never ranks above a model with data.
New! (grace period): A model less than 7 days old shows New! instead of a score and holds no place in the rankings. Its score is computed from day one, but a model one day old grows faster than anything, so without the wait the week's #1 would always be whatever came out yesterday rather than what held up for a week.
Trend: How a model's growth over a period compares with the typical model's growth over the same period. Faster than typical reads as a rise, slower as a fall.
Typical model: The middle of the catalog: half the models with a figure grew faster over the period, half slower.

Similar Models

Pareto Latest

TEXT / LLM

Qwen3.5 Plus 2026 02 15

TEXT / LLM

Qwen3 VL 8B Thinking

TEXT / LLM

Mistral Small 3.1 24B

TEXT / LLM

AI Models for Image, Video & Text Generation

MiniMax M2.1 is a text model for the Text / LLM family listed on ModelTrends.ai, a read-only catalog of open source AI image, video and text models. Compare it with other Text / LLM models and check its heat score to see how it is trending.

More Text / LLM models llm minimax reasoning coding agents multilingual local

Browse by Model Family