MiniMax M3 vs M2.7: Context, Price & License

Last verified: July 18, 2026.

The MiniMax M3 vs M2.7 decision favors M3 for most new integrations when you need image or video input, more than 204,800 tokens of combined context, controllable reasoning, or MiniMax’s M-series model positioned for long-running coding agents. MiniMax M2.7 remains useful for established text-only workflows, reproducible M2.7 evaluations, and the hosted MiniMax-M2.7-highspeed option.

The choice is not simply “newer model wins.” At the standard hosted API tier, M3 with no more than 512K input tokens has the same listed input and output rates as standard M2.7. The practical differences are context capacity, multimodal input, reasoning controls, architecture, high-speed availability, and materially different terms for self-hosting the weights.

Independent comparison: MiniMax-AI.chat is not affiliated with MiniMax. Benchmark figures below are attributed to MiniMax and are not tests run by this website. Prices, access rules, and license terms are dated snapshots; verify the official sources before purchasing capacity or deploying commercially.

MiniMax M3 vs M2.7: quick comparison

Area	MiniMax M3	MiniMax M2.7
Release date	June 1, 2026	March 18, 2026
Hosted model ID	`MiniMax-M3`	`MiniMax-M2.7`; faster hosted variant `MiniMax-M2.7-highspeed`
Combined context window	1,000,000 tokens	204,800 tokens
Documented output setting	131,072 recommended; 524,288 hard maximum, subject to the combined context budget	65,536 recommended; 204,800 hard maximum, subject to the combined context budget
Compatible inputs	Text, images, video, tool calls, tool results, and thinking content	Text, tool calls, tool results, and thinking content; no image or video input through the compatible text APIs
Output modality	Text	Text
Reasoning behavior	Can be enabled, adaptive, or disabled; endpoint defaults differ	Thinking cannot be disabled in the documented compatible APIs
Architecture	Multimodal MoE with MiniMax Sparse Attention; approximately 428B total and 23B activated parameters	Text-focused, full-attention MoE; Hugging Face reports 229B total parameters
Public weights	`MiniMaxAI/MiniMax-M3`	`MiniMaxAI/MiniMax-M2.7`
Standard API input / output price	$0.30 / $1.20 per million tokens at up to 512K input; $0.60 / $2.40 above 512K input	$0.30 / $1.20 per million tokens
Weight-license commercial path	Conditional commercial use: attribution plus notice below the stated revenue threshold; prior authorization above it	Prior written MiniMax authorization required for any commercial use of the weights
Primary reason to choose	Long, multimodal, agentic, or repository-scale work	Existing text-only systems, M2.7 behavior, or the highspeed API variant

For deep specifications rather than a head-to-head decision, use the separate MiniMax M3 model guide and MiniMax M2.7 model guide. This page concentrates on the differences that can change an architecture or purchasing decision.

Release sequence and product positioning

MiniMax released M2.7 on March 18, 2026 and described it as its first model to participate deeply in parts of its own development process. The phrase “self-improvement” refers to an internal development workflow in which the model helped maintain memory, create skills, analyze failed experiments, and revise an agent scaffold. It does not mean that an ordinary M2.7 API request retrains or rewrites the production model.^[2]

M3 followed on June 1, 2026. MiniMax positioned it around three combined capabilities: coding and agent work, a one-million-token context window, and native image/video understanding. MiniMax Code also moved to M3, and MiniMax’s migration material added M3 access to the revised Token Plan while retaining M2.7 in the shared model pool.^[1]^[6]

M2.7 has not disappeared. The API documentation verified for this comparison still lists MiniMax-M2.7 and MiniMax-M2.7-highspeed. That makes M2.7 a supported comparison target, not merely a historical checkpoint. A team can keep it while validating M3, but a new application should have a task-specific reason to select M2.7 rather than inheriting it accidentally.

Coding and agent workflows

Both models are designed for more than single-turn code generation. They can participate in an agent loop that reads repository files, calls tools, runs tests, inspects errors, and revises a patch. In either case, the model does not obtain shell, browser, database, or desktop access by itself. The surrounding agent harness supplies those capabilities and must enforce permissions.

Where M3 changes the workflow

Larger working history: M3 can retain substantially more code, specifications, tool results, and conversation history in one combined context budget.
Visual evidence: screenshots, diagrams, interface recordings, and video can be supplied alongside text through supported compatible requests.
Controllable reasoning: an application can use adaptive thinking for difficult work and disable it for low-latency extraction or completion.
Long-horizon positioning: MiniMax’s M3 release emphasizes extended tool sessions, repository-scale work, and desktop-oriented agents.

Where M2.7 can still be the practical choice

Your production prompts, parsers, tool loops, and regression suite have already been tuned to M2.7 behavior.
The workload is text-only and reliably fits within 204,800 combined tokens.
You need the separately priced MiniMax-M2.7-highspeed model ID and have measured that its lower latency offsets its higher token rates.
You must reproduce an earlier M2.7 benchmark, customer result, or regulated validation rather than silently changing the underlying model.

For the official coding product rather than a custom harness, see the MiniMax Code setup guide. For a route chooser covering Code, Agent, Hub, Hailuo, Audio, and the API, use How to Use MiniMax AI.

Context window and maximum output

The API catalog lists a 1,000,000-token context window for M3 and 204,800 tokens for both M2.7 variants. MiniMax defines the context maximum as the total of input and output tokens, not an input allowance followed by a separate full-size answer.^[3]

system instructions
+ conversation history
+ tool definitions
+ tool calls and results
+ attached text / image / video tokens
+ model reasoning
+ final answer
≤ the model's context window

MiniMax’s Anthropic-compatible reference recommends max_tokens: 131072 for M3 and permits a hard setting of 524,288. For M2.x models, it recommends 65,536 and documents 204,800 as the hard maximum. These are output settings, not extra tokens beyond the context window. A 900K-token M3 input cannot also return a 524K-token completion.^[4]

Five times the context does not mean five times the useful evidence

M3’s nominal window is about 4.88 times M2.7’s. It can fit workloads M2.7 cannot accept in one request, but a larger ceiling does not guarantee perfect retrieval or reasoning across every token. Duplicate files, generated assets, irrelevant logs, and stale documentation can add cost and distract the model.

Remove binaries, build output, vendored dependencies, and duplicate documents.
Label files and evidence boundaries clearly.
Ask for file, section, line, or quoted-evidence references.
Test retrieval from the beginning, middle, and end of a long prompt.
Compare full-context prompting with retrieval and staged summarization.

If your cleaned task package fits inside roughly 180K tokens with enough space for tools and output, M2.7’s smaller window may not be a limiting factor. If the job genuinely requires hundreds of thousands of source tokens, M3 is the viable first-party option of the two.

Multimodality and reasoning controls

M3 accepts text, image, and video input and produces text. M2.7 is text-focused through MiniMax’s OpenAI- and Anthropic-compatible APIs; its compatible message schema supports text, tools, tool results, and thinking content but not image or video blocks. This is a functional boundary, not a small benchmark difference.^[4]

Task	Better fit	Reason
Review a text-only pull request and test logs	Evaluate both	The task may fit either model; compare correctness, retries, latency, and spend.
Compare a web implementation with a screenshot	M3	Supported visual input can be included in the same model request.
Analyze a screen-recorded failure path	M3	M3 supports video input; M2.7’s compatible text interface does not.
Low-latency classification with reasoning disabled	M3	M3 exposes a documented no-thinking mode; M2.x thinking cannot be disabled.
Preserve behavior in an existing M2.7 agent	M2.7	A model switch can change tool selection, verbosity, reasoning, and parsing.

Set M3 reasoning behavior explicitly. MiniMax documents different omission defaults across compatible endpoints: Anthropic Messages disables M3 thinking when the field is omitted, OpenAI Chat Completions uses adaptive thinking by default, and the Responses API disables reasoning when its reasoning object is omitted. Relying on omission can produce accidental latency and cost differences during migration.^[12]^[13]

In multi-step tool use, preserve the complete assistant response in history, including thinking or reasoning fields and tool-call blocks. Validate every tool argument, restrict permissions, cap steps and spend, and require approval before destructive actions. Model capability is not authorization to execute a command.^[15]

Architecture, model size, and public weights

Both models use a Mixture-of-Experts design, but M3 is not a simple weight update to M2.7. The official M3 card reports approximately 428 billion total parameters and about 23 billion activated per token. The M2.7 card reports 229 billion total parameters; its public configuration identifies 256 routed experts with eight selected per token.^[7]^[8]

Full attention in M2.7 versus MSA in M3

M2.7 belongs to MiniMax’s full-attention M2 architecture line. M3 introduces MiniMax Sparse Attention (MSA), a block-sparse mechanism built on Grouped Query Attention. A lightweight index branch selects relevant key-value blocks for each attention group, and the main branch performs attention over those selected blocks rather than every preceding token.^[14]

The MSA paper reports large compute and wall-clock improvements at one-million-token sequences, but its controlled architecture experiments use a 109B-parameter research model. Those results support the attention design; they are not an independent guarantee for every 428B M3 deployment, GPU, quantization, batch size, or API provider.^[9]

Open-weight does not mean laptop-sized

MiniMax publishes both checkpoints, and the model cards list deployment routes including vLLM, SGLang, and Transformers. That makes private hosting and inspection possible, subject to each license. It does not make either model a routine laptop deployment. Sparse activation reduces per-token computation, but the full expert weights still need storage and a serving strategy. M3 has almost twice M2.7’s total parameter count and adds multimodal components.

Hosted and self-hosted results can diverge because of quantization, precision, kernels, context length, chat templates, tool parsers, sampling, parallelism, and vision/video support. Treat an engine or quantization change as a model-system change and rerun the evaluation suite.

If M3 is your self-hosting candidate, use the MiniMax M3 local deployment guide to size BF16 and quantized checkpoints, RAM, VRAM, long-context cache, vLLM, and SGLang before selecting hardware.

What the official benchmark claims do—and do not—show

MiniMax reports 59.0% for M3 on SWE-Bench Pro and 56.22% for M2.7 on SWE-Pro. The official Hugging Face evaluation metadata also displays approximately 59.0 and 56.2 for SWE-Bench Pro. That is a useful directional signal, but it is still provider-published evidence rather than a controlled evaluation performed by this website.^[1]^[2]^[7]^[8]

Benchmark	M3 result reported by MiniMax	M2.7 result reported by MiniMax	Comparison warning
SWE-Bench Pro / SWE-Pro	59.0%	56.22%	The roughly 2.78-point difference is vendor-reported; verify scaffold and run settings before treating it as an A/B result.
Terminal-Bench	66.0% on Terminal-Bench 2.1	57.0% on Terminal Bench 2	The benchmark versions differ, so the raw values are not directly interchangeable.
BrowseComp	83.5 on the official M3 product page^[16]	No directly paired M2.7 value on the cited product page	Do not invent a head-to-head delta where the same table does not provide one.
Long-running demonstrations	MiniMax reports multi-hour paper reproduction, model-training, and CUDA optimization tasks	MiniMax reports scaffold self-optimization and agent/productivity evaluations	These are first-party case studies, not independent replications.

Benchmark outcomes can change with the agent scaffold, tool access, repository image, timeout, retry policy, pass count, judge model, prompt, and compute budget. The correct purchasing question is not which model has the larger headline score. It is which model completes your tasks correctly at an acceptable cost and with the least human correction.

MiniMax M3 vs M2.7 API price

The table below is a standard-service pay-as-you-go snapshot verified on July 18, 2026. Rates are in U.S. dollars per one million tokens. MiniMax’s page labels the displayed M3 rates as a permanent 50% reduction from struck-through list rates, but the dated displayed amount—not the marketing label—should be used for this comparison.^[5]

Hosted model / input band	Input	Output	Prompt-cache read	Prompt-cache write
`MiniMax-M3`, ≤512K input tokens	$0.30/M	$1.20/M	$0.06/M	Not listed in the M3 row
`MiniMax-M3`, >512K input tokens	$0.60/M	$2.40/M	$0.12/M	Not listed in the M3 row
`MiniMax-M2.7`	$0.30/M	$1.20/M	$0.06/M	$0.375/M
`MiniMax-M2.7-highspeed`	$0.60/M	$2.40/M	$0.06/M	$0.375/M

Key pricing conclusion: M3 and standard M2.7 have the same ordinary input, output, and cache-read rates while M3 input remains at or below 512K. M3 becomes twice as expensive per input, output, and cache-read token when the request crosses the 512K input threshold. M2.7-highspeed doubles ordinary input/output rates but retains M2.7’s 204,800-token context window.

Two transparent cost examples

Assume standard service, no cache hits, no retries, and text token billing only.

Example A — 180K input + 10K output
M3 (≤512K band): 0.18 × $0.30 + 0.01 × $1.20 = $0.066
M2.7 standard:    0.18 × $0.30 + 0.01 × $1.20 = $0.066
M2.7-highspeed:   0.18 × $0.60 + 0.01 × $2.40 = $0.132

Example B — 600K input + 20K output
M3 (>512K band):  0.60 × $0.60 + 0.02 × $2.40 = $0.408
M2.7:              request exceeds its 204,800-token context window

These are calculations, not observed invoices. Images, video, reasoning, tool history, cache behavior, failed calls, service tier, and provider markup can change the total. Compare cost per successful task, not just the advertised rate per million tokens. See the broader MiniMax pricing guide for pay-as-you-go keys, Subscription Keys, Credits, and plan rules.

Token Plan positioning

MiniMax’s M3-era migration guide says M3 access is included for migrated plans and shares the quota pool with M2.7. It also describes quota protection and compensation for affected subscriptions. A Token Plan is not the same billing surface as a pay-as-you-go API account, and a Subscription Key is separate from an ordinary API Key.^[6]

Do not convert a per-token pay-as-you-go rate into a guaranteed Token Plan allowance. Plan tiers, quota windows, shared modalities, concurrency, and migration compensation are separate rules. Check the billing console attached to the exact account before capacity planning.

M3 vs M2.7 license: the decisive self-hosting difference

Both models have downloadable weights, but neither should be described as unrestricted, permissively licensed open source. “Open-weight under a custom license” is the more accurate wording. Their commercial paths are different enough to affect model selection.

License point	MiniMax M3	MiniMax M2.7
License name	MiniMax Community License	Non-Commercial License
Non-commercial use	Permitted subject to the license and retained notices	Permitted for personal use and non-profit research, subject to the license and prohibited-use policy
Commercial attribution	Prominently display “Built with MiniMax M3”	Prominently display “Built with MiniMax M2.7” if commercially authorized
Commercial use at or below the stated revenue threshold	The license requires a one-time notice to MiniMax; it states the threshold in relation to products or services generating no more than $20M in yearly revenue	Prior written authorization is required; the license provides no small-business revenue exception
Commercial use above the M3 threshold	Prior written MiniMax authorization is required	Prior written MiniMax authorization is required for commercial use at any size
Prohibited uses	Yes	Yes

The M3 terms therefore provide a defined notice-based route for some commercial deployments that M2.7’s weight license does not. That does not make M3 equivalent to Apache 2.0 or MIT, and it does not eliminate attribution, notification, revenue, prohibited-use, or legal-review obligations.^[10]^[11]

Keep the legal surfaces separate. Downloading, modifying, or serving the public checkpoint invokes the model-weight license. Calling MiniMax’s hosted API is governed by the applicable platform and paid-service terms. Paying API fees does not grant permission to self-host the weights commercially, and a weight-license summary does not replace the hosted service agreement.

This section summarizes source text and is not legal advice. Keep a dated copy of the exact license evaluated, check the checkpoint and derivative license, and obtain qualified legal review before commercial deployment or redistribution.

Decision matrix: which model should you choose?

Your requirement	Start with	Why
New coding-agent integration	M3	Same base token rate as M2.7 within 512K input, larger context, multimodal input, and controllable reasoning
Repository or document package above 204,800 combined tokens	M3	M2.7 cannot accept the full package in one request
Screenshot, diagram, or video understanding	M3	M3 supports these input modalities through the documented compatible API
Stable production system already validated on M2.7	Keep M2.7 during evaluation	Model replacement can change output shape, tool behavior, reasoning, latency, and regressions
Documented faster hosted M2.7 variant	M2.7-highspeed	MiniMax offers a dedicated highspeed model ID; validate whether the doubled input/output rate is justified
Self-hosted commercial product below M3’s stated $20M product/service threshold	M3 may have the clearer path	M3 specifies attribution plus notice; M2.7 requires prior written authorization for any commercial use
Permissive OSI-style license required by policy	Neither without legal approval	Both use custom licenses with commercial conditions and prohibited uses
Typical consumer laptop deployment	Neither base checkpoint	Both are very large MoE models; M3 is substantially larger and multimodal
Exact reproduction of an M2.7 result	M2.7	Changing the model invalidates an exact reproduction claim

If you are choosing an API for a new system and have no M2.7 compatibility constraint, evaluate M3 first. If you operate an established M2.7 workflow, run both behind the same interface and migrate only after M3 passes the regression, safety, latency, and cost gates.

How to migrate from MiniMax M2.7 to M3 safely

The base URLs do not need to change when both models are called through the same MiniMax compatible route. The minimum code change may be only the model ID, but a production migration is larger than a string replacement. See the MiniMax API guide for endpoint setup and SDK patterns.

1. Make the model configurable

# Before
MINIMAX_MODEL=MiniMax-M2.7

# Candidate
MINIMAX_MODEL=MiniMax-M3

Keep an immediate M2.7 fallback while validation is in progress. Do not scatter a model ID through application code, tests, job definitions, and agent configuration.

2. Set M3 thinking explicitly

Choose enabled/adaptive or disabled based on the task and exact API format. Test both difficult reasoning work and simple low-latency calls. Do not assume M2.7’s always-on thinking behavior maps automatically to M3 or that every compatible endpoint has the same default.

3. Update message and output parsers

Preserve complete thinking, text, tool-use, and tool-result blocks.
Confirm that structured output and tool arguments still pass schema validation.
Add image/video handling only after validating supported content types, upload routes, and size limits.
Cap output below both the configured maximum and the remaining combined context.

4. Recalculate context and cost guards

Raise the application context limit only when a task needs it. Add an alert before 512K input because crossing that M3 threshold changes all three listed standard rates for the call. Keep retrieval, deduplication, and staged processing available rather than treating 1M as a target prompt size.

5. Run a controlled A/B evaluation

Use the same repository snapshot, prompts, tools, permissions, sandbox, timeout, and retry policy.
Include bug repair, cross-file refactoring, test generation, repository navigation, ambiguous requirements, and tool failures.
Require tests or deterministic checks to pass; do not score code by appearance alone.
Record task success, invalid tool calls, regressions, input/output/cache tokens, time to first token, total wall time, retries, and human correction time.
Calculate cost per accepted task, not cost per response.

6. Roll out gradually

Start with low-risk traffic, observe tool and parser failures, and retain a model-level rollback. Revalidate privacy, rate limits, budget alarms, and authorization gates. A successful benchmark run is not a production approval.

A practical scorecard for your own comparison

Metric	How to measure	Why it matters
Task success	Automated tests plus reviewer rubric	Primary evidence that the model solved the job
Regression rate	Previously passing tests broken by the patch	Measures collateral damage
Tool reliability	Valid calls / total calls; recovery from failures	Agent quality depends on execution, not prose
Human correction	Minutes and edits required before acceptance	A slightly slower model may reduce total labor
End-to-end latency	Task start to accepted result	Includes tools, retries, and reasoning
Total cost	Input + output + cache + retry + tool/provider costs	Per-token price alone can mislead
Long-context retrieval	Known evidence placed at multiple prompt positions	Tests whether extra context is usable
Safety	Unauthorized action attempts, secret exposure, unsafe commands	Strong coding ability increases the importance of controls

Run several trials per task because sampling and tool environments introduce variance. Publish your scaffold, versions, prompts, pass count, and scoring rules if you later describe the result as an original benchmark.

Frequently asked questions

Is MiniMax M3 better than M2.7?

M3 has the stronger feature set: a 1M combined context window, image and video input, controllable reasoning, a newer sparse-attention architecture, and higher MiniMax-reported coding scores. M2.7 can still be preferable for validated text-only systems, exact result reproduction, or use of its highspeed hosted variant. “Better” should be decided by task success, correction time, latency, safety, and total cost on the same workload.

Does M3 cost more than M2.7?

Not at the standard pay-as-you-go tier while M3 input is at or below 512K tokens. Both list $0.30 per million input tokens and $1.20 per million output tokens, with $0.06 per million cache-read tokens. Above 512K M3 input, the displayed M3 rates double. These prices were verified on July 18, 2026.

Is M2.7 deprecated after the M3 release?

No deprecation is stated in the official pages used for this comparison. MiniMax’s API catalog still documents MiniMax-M2.7 and MiniMax-M2.7-highspeed. Support can change, so monitor release notes before making a long-term commitment.

Can M2.7 accept images or video?

Not through MiniMax’s documented compatible text APIs. M2.x models accept text and tool-call content there, while M3 supports text, image, and video input. Both produce text output in this comparison.

Does M3 provide five times more context?

M3’s 1,000,000-token ceiling is about 4.88 times M2.7’s 204,800-token ceiling. The total includes input and output. Capacity does not guarantee perfect retrieval, and M3’s standard API rates increase when input exceeds 512K tokens.

Can I switch by changing only the model ID?

A basic text call may work after changing MiniMax-M2.7 to MiniMax-M3 on the same compatible endpoint. Production migration also requires explicit reasoning settings, parser checks, context and cost guards, tool-loop regression tests, and staged rollout.

Are MiniMax M3 and M2.7 open source?

The precise description is open-weight under custom licenses. M3 uses the MiniMax Community License; M2.7 uses a Non-Commercial License. Both allow downloading and permitted self-hosting, but neither is an unrestricted MIT- or Apache-style model-weight release.

Which license is easier for commercial self-hosting?

M3 provides a conditional notice-based route when the relevant products or services do not exceed the license’s $20M yearly-revenue threshold, with attribution required. M2.7 requires prior written authorization for any commercial use of the weights. Review the full license and obtain legal advice; hosted API terms are a separate contract.

Official sources used

MiniMax M3 release report — release date, product positioning, MSA claims, multimodality, and MiniMax-reported benchmarks.
MiniMax M2.7 release report — release date, self-improvement framing, coding, agent, and productivity positioning.
MiniMax model invocation guide — hosted model IDs, context windows, compatible endpoints, and supported-model status.
MiniMax Anthropic-compatible Messages reference — input types, output limits, thinking behavior, and content blocks.
MiniMax pay-as-you-go pricing — M3 input bands and M2.7 standard/highspeed token rates.
MiniMax Token Plan migration guide — M3 access, shared M2.7 quota, and migration treatment.
Official MiniMax M3 model card — parameter counts, modalities, reasoning modes, weights, and deployment routes.
Official MiniMax M2.7 model card — parameter count, weights, model configuration, evaluation metadata, and deployment routes.
MiniMax Sparse Attention paper — MSA design and the scope of its 109B experimental results.
MiniMax M3 Community License — commercial notice, attribution, revenue threshold, authorization, and prohibited uses.
MiniMax M2.7 Non-Commercial License — commercial authorization, attribution, permitted free uses, and prohibited uses.
MiniMax OpenAI-compatible Chat Completions reference — M3 adaptive-thinking default and M2.x reasoning behavior.
MiniMax Responses API reference — M3 reasoning control and the no-reasoning omission default.
MiniMax M2 full-attention engineering note — the M2 family’s full-attention design choice and limitations MiniMax identified in earlier efficient-attention experiments.
MiniMax M3 tool-use and interleaved-thinking guide — preserving complete reasoning and tool-call content across turns.
Official MiniMax M3 product page — product capabilities and MiniMax-reported BrowseComp result.

Scope: This is a source-based comparison, not a hands-on benchmark or legal opinion. Facts and displayed prices were checked against the linked sources on July 18, 2026.

MiniMax M3 vs M2.7: Coding, Context, Price, and License