Skip to main content
BenchLM
Data

MiniMax M3 vs Qwen 3.6 Max (preview)

Updated October 14, 2026. Rank cannot separate these two. Price, access, and your workload decide. Public scores include evidence status and uncertainty.

Share or export
Share on XLinkedInAPI/MCP

Decision reading

Qwen 3.6 Max (preview) has the higher public point estimate, 55.25 versus 55. Their conditional score ranges overlap. These ranges do not establish rank confidence. 3 results are shared. Category rows resting on Estimated evidence or different benchmark sets are marked directional and do not name a winner.

Model A
MiniMax logo
MiniMax M3

MiniMax

55/100

Supported · Public rank #68

90% interval 45.6–64.4

Model B
Alibaba logo
Qwen 3.6 Max (preview)

Alibaba

55.25/100

Estimated · Public rank #67

Conditional range 45.5–65.0

Shared results
3
MiniMax M3 only
24
Qwen 3.6 Max (preview) only
5
Like-for-like categories
0 / 8
Supported: MiniMax M3 · Estimated: Qwen 3.6 Max (preview). Conditional ranges do not establish rank confidence.How the comparison works

Which one for your work

Recommendations appear only when a shared evidence basis or an explicit operating constraint supports the call. Secondary and unsupported use cases stay disclosed below the initial list.

  • Long documents

    Prompts that approach the documented context limit

    MiniMax M3

    MiniMax M3 has the larger documented context window.

    Confidence: documented
Show secondary and unsupported calls

  • Coding work

    Code generation, repair, and software-engineering tasks

    Not enough matched evidence

    Qwen 3.6 Max (preview) is scored on Estimated evidence for coding, so the reading is directional rather than like-for-like.

    Confidence: limited

  • Agentic work

    Tool use, computer use, and multi-step task completion

    Not enough matched evidence

    Qwen 3.6 Max (preview) is not ranked on the public lane for agentic, so no winner is named for agentic.

    Confidence: limited

  • Chat turn cost

    1K fresh input + 500 output tokens

    Not enough matched evidence

    A complete comparable API-rate estimate is not available for both models.

    Confidence: listed-rates

  • Cache-heavy agent loop cost

    200K cached + 20K fresh input + 10K output tokens

    Not enough matched evidence

    A complete comparable API-rate estimate is not available for both models.

    Confidence: listed-rates

  • Repository review cost

    50K fresh input + 3K output tokens

    Not enough matched evidence

    A complete comparable API-rate estimate is not available for both models.

    Confidence: listed-rates

Which one for a specific job

Choose a job from the LLM Selector's task catalog to see the category row it rests on, under the same basis rules as the table below. A directional row stays directional; choosing a job never creates a winner.

The same task catalog as the LLM Selector. Each job names the evidence surface it rests on; nothing here adds a new score.

36.5MiniMax M342.4Qwen 3.6 Max (preview)

Directional only · BenchAlign v5.8

Qwen 3.6 Max (preview) has the higher coding point estimate. Conditional score ranges do not establish rank confidence.

Coding scores combine specific tasks and setups. Match the editor, harness, and effort to your workflow.

Open the coding leaderboardShortlist models for this job
Same basis rules as the category table below

What is actually comparable

Shared results can support a head-to-head reading. Results present for only one model describe coverage, not superiority.

1 category rests on Estimated evidence or different evidence sets. Those rows remain visible for coverage context but do not name a winner.

Shape of the matched evidence

Only shared public evidence is shown. Sparse evidence stays a ruled list rather than being closed into a radar shape.

Too few matched category axes support a radar. The ruled list below shows only shared benchmark results; positions use each benchmark’s normalized display scale when available.

Bars run 0–100 on each benchmark’s normalized display scale

Category results, on a stated basis

Each row shows the public-lane category score for both models: the BenchAlign v5.8 lane for agentic, coding, and knowledge, and the weighted public rows elsewhere, with the same rank each model page reports. A row is like-for-like only when both scores rest on Supported evidence or the same weighted set. Directional and not-comparable rows remain visible, but they never receive a winner.

Coding

Directional only
MiniMax M3
36.5
Supported · #73/149
Qwen 3.6 Max (preview)
42.4
Estimated · #59/149
Basis
BenchAlign v5.8 lane · 10 vs 5 public rows
Reading
Directional only

Agentic

Not comparable
MiniMax M3
39.6
Supported · #66/125
Qwen 3.6 Max (preview)
Not ranked
Basis
BenchAlign v5.8 lane · 9 vs 2 public rows
Reading
Not comparable

Reasoning

Not comparable
MiniMax M3
82.9
#5/29
Qwen 3.6 Max (preview)
Not ranked
Basis
Provisional lane · 0 vs 0 weighted rows
Reading
Not comparable

Multimodal

Not comparable
MiniMax M3
61.5
#43/54
Qwen 3.6 Max (preview)
Not ranked
Basis
Provisional lane · 2 vs 0 weighted rows
Reading
Not comparable

Knowledge

Not comparable
MiniMax M3
49.6
Supported · #80/181
Qwen 3.6 Max (preview)
Not ranked
Basis
BenchAlign v5.8 lane · 2 vs 1 public rows
Reading
Not comparable

Multilingual

Not comparable
MiniMax M3
Not ranked
Qwen 3.6 Max (preview)
Not ranked
Basis
Provisional lane · 0 vs 0 weighted rows
Reading
Not comparable

Instruction following

Not comparable
MiniMax M3
93.8
#5/144
Qwen 3.6 Max (preview)
Not ranked
Basis
Provisional lane · 0 vs 0 weighted rows
Reading
Not comparable

Math

Not comparable
MiniMax M3
Not ranked
Qwen 3.6 Max (preview)
Not ranked
Basis
Provisional lane · 1 vs 0 weighted rows
Reading
Not comparable

Ranks count the models scored in each category’s lane, so the agentic, coding, and knowledge denominators (BenchAlign v5.8) differ from the provisional-lane categories. Unranked scores sit on the provisional lane’s scale but never name a winner. Like-for-like rows within 0.5 points read as a practical tie.

Supported evidence per lane · bars run 0–100Methodology

What each workload costs

Three fixed token mixes turn per-token rates into comparable decisions. Each scenario states context fit and whether cached input had to fall back to the published list-input rate.

Chat turn

1K fresh input + 500 output tokens

MiniMax M3
$0.0009
Fits in one request
Qwen 3.6 Max (preview)
API rate not published
Fits in one request

Qwen 3.6 Max (preview) has no comparable published API token rate.

Repository review

50K fresh input + 3K output tokens

MiniMax M3
$0.0186
Fits in one request
Qwen 3.6 Max (preview)
API rate not published
Fits in one request

Qwen 3.6 Max (preview) has no comparable published API token rate.

Cache-heavy agent loop

200K cached + 20K fresh input + 10K output tokens

MiniMax M3
$0.03
Fits in one request
Qwen 3.6 Max (preview)
API rate not published
Fits in one request
Cached-input rate unavailable

Qwen 3.6 Max (preview) has no comparable published API token rate.

Cached input falls back to the list input rate only where a cached rate is unpublished

Specification differences

Sourced differences are shown directly. Missing facts stay explicit instead of being inferred from a model name or family.

Context window

Maximum documented context; output-token limits may be lower.

MiniMax M3

1M

Qwen 3.6 Max (preview)

256K

API model ID

MiniMax M3

Not sourced

Qwen 3.6 Max (preview)

Not sourced

Cached-input rate

A missing cached-input rate falls back to the listed input rate only in the stated workload estimate.

MiniMax M3

$0.06 per 1M cached input tokens

Qwen 3.6 Max (preview)

No comparable hosted API rate

Documented inputs

MiniMax M3

Not sourced

Qwen 3.6 Max (preview)

Not sourced

Documented outputs

MiniMax M3

Not sourced

Qwen 3.6 Max (preview)

Not sourced

Provider availability

MiniMax M3

Not sourced

Qwen 3.6 Max (preview)

Not sourced

Reasoning profile

MiniMax M3

Non-Reasoning

Qwen 3.6 Max (preview)

Reasoning

Weight access

MiniMax M3

Open Weight

Qwen 3.6 Max (preview)

Proprietary

License

MiniMax M3

Open Weight

Qwen 3.6 Max (preview)

Proprietary

Release date

MiniMax M3

2026-06-01

Qwen 3.6 Max (preview)

2026-04-20

If you already use one of these models

Deployment change
The models list different providers, so authentication, endpoint behavior, limits, and feature support may change.
Quality signal
Qwen 3.6 Max (preview) has the higher public point estimate, 55.25 versus 55. Their conditional score ranges overlap. These ranges do not establish rank confidence.
Workload cost
A complete comparable API-rate estimate is not available for both models.
Context tradeoff
MiniMax M3 has the larger documented window (1M).
Run the same representative tasks against both endpoints before changing production traffic.

Questions

Which is better, MiniMax M3 or Qwen 3.6 Max (preview)?

Qwen 3.6 Max (preview) has the higher public point estimate, 55.25 versus 55. Their conditional score ranges overlap. These ranges do not establish rank confidence. The page therefore keeps the decision tied to the specific documented workload.

Which is better for coding, MiniMax M3 or Qwen 3.6 Max (preview)?

Qwen 3.6 Max (preview) scores higher for coding on the public lane, 42.4 to 36.5. Qwen 3.6 Max (preview) is scored on Estimated evidence for coding, so the reading is directional rather than like-for-like. BenchLM does not name a winner for coding from a directional reading; read the shared benchmark rows directly and test the models on the same task set.

Which is better for agentic tasks, MiniMax M3 or Qwen 3.6 Max (preview)?

Qwen 3.6 Max (preview) is not ranked on the public lane for agentic tasks, so no winner is named for agentic tasks.

Which costs less, MiniMax M3 or Qwen 3.6 Max (preview)?

Both models do not have comparable published API token rates, so this page does not name a universal price winner.

Which has the larger context window, MiniMax M3 or Qwen 3.6 Max (preview)?

MiniMax M3 has the larger documented context window: 1M, compared with 256K.

Benchmark evidence

The full public result ledger is available for audit without forcing a wide desktop table onto a phone.

Browse raw public benchmark evidence32 rows

Agentic

  • Terminal-Bench 2.1

    MiniMax M366.0%
    Source
    Qwen 3.6 Max (preview)—

    Not directly comparable

  • BrowseComp

    MiniMax M383.5%
    Source
    Qwen 3.6 Max (preview)—

    Not directly comparable

  • OSWorld-Verified

    MiniMax M370.1%
    Source
    Qwen 3.6 Max (preview)—

    Not directly comparable

  • MCP Atlas

    MiniMax M374.2%
    Source
    Qwen 3.6 Max (preview)—

    Not directly comparable

  • Claw-Eval

    MiniMax M374.5%
    Source
    Qwen 3.6 Max (preview)—

    Not directly comparable

  • BankerToolBench

    MiniMax M376.1%
    Source
    Qwen 3.6 Max (preview)—

    Not directly comparable

  • ResearchClawBench

    MiniMax M319.8%
    Source
    Qwen 3.6 Max (preview)—

    Not directly comparable

  • OSWorld 2.0

    MiniMax M34.6%
    Source
    Qwen 3.6 Max (preview)—

    Not directly comparable

  • Terminal-Bench 2.1 (Vals)

    MiniMax M353.6%
    Source
    Qwen 3.6 Max (preview)—

    Not directly comparable

  • Terminal-Bench 2.0

    MiniMax M3—
    Qwen 3.6 Max (preview)65.4%
    Source

    Not directly comparable

  • QwenClawBench

    MiniMax M3—
    Qwen 3.6 Max (preview)59.0%
    Source

    Not directly comparable

Coding

  • SWE-bench Verified

    MiniMax M380.5%
    Source
    Qwen 3.6 Max (preview)—

    Not directly comparable

  • SWE-bench Pro

    MiniMax M359%
    Source
    Qwen 3.6 Max (preview)57.3%
    Source

    MiniMax M3 leads this result

  • Terminal-Bench 2.1

    MiniMax M366.0%
    Source
    Qwen 3.6 Max (preview)—

    Not directly comparable

  • NL2Repo

    MiniMax M342.1%
    Source
    Qwen 3.6 Max (preview)42.9%
    Source

    Qwen 3.6 Max (preview) leads this result

  • VIBE V2

    MiniMax M350.1%
    Source
    Qwen 3.6 Max (preview)—

    Not directly comparable

  • SVG-Bench

    MiniMax M363.7%
    Source
    Qwen 3.6 Max (preview)—

    Not directly comparable

  • KernelBench Hard

    MiniMax M328.8%
    Source
    Qwen 3.6 Max (preview)—

    Not directly comparable

  • OpenHarmony Bench

    MiniMax M348.4%
    Source
    Qwen 3.6 Max (preview)—

    Not directly comparable

  • LiveCodeBench (Vals)

    MiniMax M382.2%
    Source
    Qwen 3.6 Max (preview)—

    Not directly comparable

  • SWE-bench (Vals)

    MiniMax M375.0%
    Source
    Qwen 3.6 Max (preview)72.8%
    Source

    MiniMax M3 leads this result

  • SciCode

    MiniMax M3—
    Qwen 3.6 Max (preview)47%
    Source

    Not directly comparable

  • Terminal-Bench 2.0

    MiniMax M3—
    Qwen 3.6 Max (preview)65.4%
    Source

    Not directly comparable

Multimodal

  • OfficeQA Pro

    MiniMax M345.1%
    Source
    Qwen 3.6 Max (preview)—

    Not directly comparable

  • OmniDocBench 1.5

    MiniMax M391.6%
    Source
    Qwen 3.6 Max (preview)—

    Not directly comparable

  • MMMU-Pro

    MiniMax M378.1%
    Source
    Qwen 3.6 Max (preview)—

    Not directly comparable

  • VideoMMMU

    MiniMax M384.6%
    Source
    Qwen 3.6 Max (preview)—

    Not directly comparable

  • Video-MME (with subtitle)

    MiniMax M385.4%
    Source
    Qwen 3.6 Max (preview)—

    Not directly comparable

Knowledge

  • GPQA Diamond (Vals)

    MiniMax M392.7%
    Source
    Qwen 3.6 Max (preview)—

    Not directly comparable

  • MMLU-Pro (Vals)

    MiniMax M384.2%
    Source
    Qwen 3.6 Max (preview)—

    Not directly comparable

  • SuperGPQA

    MiniMax M3—
    Qwen 3.6 Max (preview)73.9%
    Source

    Not directly comparable

Math

  • USAMO 2026

    MiniMax M385.7%
    Source
    Qwen 3.6 Max (preview)—

    Not directly comparable

32 public results · 3 shared

Related comparisons

Compare API pricingRead the methodologyOpen the model selector

Watch MiniMax M3 vs Qwen 3.6 Max (preview)

One weekly email when material rank, price, or benchmark evidence changes make this matchup worth revisiting.

Read a sample issue

Join 5,500+ readers.

Last updated October 14, 2026