Skip to main content
BenchLM
Data

Kimi K3 vs MiniMax M2.5

Updated November 17, 2026. Rank says Kimi K3 is ahead. Price, access, and your workload can each overturn that. Public scores include evidence status and uncertainty.

Share or export
Share on XLinkedInAPI/MCP

Decision reading

The public evidence has no benchmark result shared by both models, so it does not support a quality verdict. Use the documented cost, context, and runtime rows instead. 0 results are shared. Category rows resting on Estimated evidence or different benchmark sets are marked directional and do not name a winner.

Model A
Moonshot AI logo

Moonshot AI

67/100

Supported · Public rank #20

90% interval 61.6–72.4

Model B
MiniMax logo

MiniMax

48.61/100

Supported · Public rank #100

90% interval 38.6–58.6

API price per 1M tokens

Kimi K3: $3 input / $15 output

MiniMax M2.5: $0.30 input / $1.20 output

Context window

Kimi K3: 1.05M

MiniMax M2.5: 128K

Released

Kimi K3: July 16, 2026

MiniMax M2.5: February 12, 2026

Shared results
0
Kimi K3 only
60
MiniMax M2.5 only
1
Like-for-like categories
0 / 8
Supported: Kimi K3 and MiniMax M2.5How 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

    Kimi K3

    Kimi K3 has the larger documented context window.

    Confidence: documented
  • Chat turn cost

    1K fresh input + 500 output tokens

    MiniMax M2.5

    MiniMax M2.5 has the lower estimated token cost for this stated workload. Costs use the listed standard API rates.

    Confidence: listed-rates
Show secondary and unsupported calls
  • Repository review cost

    50K fresh input + 3K output tokens

    MiniMax M2.5

    MiniMax M2.5 has the lower estimated token cost for this stated workload. Costs use the listed standard API rates.

    Confidence: listed-rates
  • Coding work

    Code generation, repair, and software-engineering tasks

    Not enough matched evidence

    MiniMax M2.5 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

    MiniMax M2.5 is scored on Estimated evidence for agentic, so the reading is directional rather than like-for-like.

    Confidence: limited
  • Cache-heavy agent loop cost

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

    Not enough matched evidence

    The page does not recommend a cost winner because at least one model cannot fit the stated workload in one request. MiniMax M2.5 does not fit this workload in one request. MiniMax M2.5 has no published cached-input rate, so cached tokens use its listed input rate.

    Confidence: rate-fallback

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.

56.3Kimi K334.9MiniMax M2.5

Directional only · BenchAlign v5.8

Kimi K3 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.

Same basis rules as the category table below

Or describe your own job

The LLM Selector reads what you need a model for and returns a shortlist, ranked in the same order the job pages publish.

Describe the task and constraints. This selector recommends models; it cannot carry out the task or access your files and accounts.

We keep what you type here, drafts included, for 180 days, and after that only the job and requirements read from it. Nothing is tied to your IP address or account. Leave out names and other personal details. Privacy

What is actually comparable

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

2 categories rest 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.

A shared-evidence shape is not available.

BenchLM does not draw a radar or infer missing axes when the matched evidence is too sparse.

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, reasoning, multimodal, 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.

Agentic

Directional only
Kimi K3
65.1
Supported · #21/140
MiniMax M2.5
30.5
Estimated · #111/140
Basis
BenchAlign v5.8 lane · 19 vs 0 public rows
Reading
Directional only

Coding

Directional only
Kimi K3
56.3
Supported · #23/163
MiniMax M2.5
34.9
Estimated · #95/163
Basis
BenchAlign v5.8 lane · 18 vs 1 public rows
Reading
Directional only

Reasoning

Not comparable
Kimi K3
65.4
Supported · #21/161
MiniMax M2.5
Not ranked
Basis
BenchAlign v5.8 lane · 2 vs 0 public rows
Reading
Not comparable

Multimodal

Not comparable
Kimi K3
66.7
Estimated · #19/67
MiniMax M2.5
Not ranked
Basis
BenchAlign v5.8 lane · 14 vs 0 public rows
Reading
Not comparable

Knowledge

Not comparable
Kimi K3
67.5
Supported · #20/181
MiniMax M2.5
Not ranked
Basis
BenchAlign v5.8 lane · 6 vs 0 public rows
Reading
Not comparable

Multilingual

Not comparable
Kimi K3
Not ranked
MiniMax M2.5
Not ranked
Basis
Provisional lane · 0 vs 0 weighted rows
Reading
Not comparable

Instruction following

Not comparable
Kimi K3
Not ranked
MiniMax M2.5
Not ranked
Basis
Provisional lane · 0 vs 0 weighted rows
Reading
Not comparable

Math

Not comparable
Kimi K3
Not ranked
MiniMax M2.5
Not ranked
Basis
Provisional lane · 0 vs 0 weighted rows
Reading
Not comparable

Ranks count the models scored in each category’s lane, so the agentic, coding, reasoning, multimodal, 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

Kimi K3
$0.0105
Fits in one request
MiniMax M2.5
$0.0009
Fits in one request

MiniMax M2.5 has the lower modeled cost

Costs use the listed standard API rates.

Repository review

50K fresh input + 3K output tokens

Kimi K3
$0.195
Fits in one request
MiniMax M2.5
$0.0186
Fits in one request

MiniMax M2.5 has the lower modeled cost

Costs use the listed standard API rates.

Cache-heavy agent loop

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

Kimi K3
$0.27
Fits in one request
MiniMax M2.5
$0.078
Does not fit in one request
Cached input priced at the published list-input rate

MiniMax M2.5 does not fit this workload in one request. MiniMax M2.5 has no published cached-input rate, so cached tokens use its listed input rate.

Your workload

Enter a month of tokens to price it at both models’ listed API rates.

Kimi K3
$60.00
per month
MiniMax M2.5
$5.40
per month

Standard listed rates, no batch or volume discounts.

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

Price against score, among all priced models

Overall score against output price per 1M tokens, one model per family. Pricing as of November 17, 2026.

Kimi K3MiniMax M2.5

20406080100$0.10$0.50$1$5$10$50$100Output price, $ per 1M tokens (log)Overall score
  • 3 cheaper models score at least as high as Kimi K3. The nearest is Gemini 4 Carbon (score 85.1, $10.00 per 1M output tokens).
  • 7 models score at least as high as MiniMax M2.5 for the same price or less. The nearest is DeepSeek V4.1 Flash 1017 (score 63.9, also $1.20 per 1M output tokens).

Gray dots are the other priced models. The dark step line is the efficiency frontier: for a model under it, another model costs no more and scores at least as high.

See MiniMax M2.5 on the full price vs performance chart

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.

Kimi K3

1.05M

MiniMax M2.5

128K

API model ID

Kimi K3

Not sourced

MiniMax M2.5

Not sourced

Cached-input rate

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

Kimi K3

$0.3 per 1M cached input tokens

MiniMax M2.5

Not published

Provider availability

Kimi K3

Not sourced

MiniMax M2.5

Not sourced

Reasoning profile

Kimi K3

Reasoning

MiniMax M2.5

Non-Reasoning

Weight access

Kimi K3

Open Weight

MiniMax M2.5

Proprietary

License

Kimi K3

Open Weight

MiniMax M2.5

Proprietary

Release date

Kimi K3

2026-07-16

MiniMax M2.5

2026-02-12

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
The public evidence has no benchmark result shared by both models, so it does not support a quality verdict.
Workload cost
Repository review: $0.195 vs $0.0186. Cache-heavy agent loop: $0.27 vs $0.078.
Context tradeoff
Kimi K3 has the larger documented window (1.05M).
Run the same representative tasks against both endpoints before changing production traffic.

Questions

Which is better, Kimi K3 or MiniMax M2.5?

The public evidence has no benchmark result shared by both models, so it does not support a quality verdict. The page therefore keeps the decision tied to the specific documented workload.

Which is better for coding, Kimi K3 or MiniMax M2.5?

Kimi K3 scores higher for coding on the public lane, 56.3 to 34.9. MiniMax M2.5 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, Kimi K3 or MiniMax M2.5?

Kimi K3 scores higher for agentic tasks on the public lane, 65.1 to 30.5. MiniMax M2.5 is scored on Estimated evidence for agentic tasks, so the reading is directional rather than like-for-like. BenchLM does not name a winner for agentic tasks from a directional reading; read the shared benchmark rows directly and test the models on the same task set.

Which costs less, Kimi K3 or MiniMax M2.5?

For the stated presets, chat costs $0.0105 on Kimi K3 and $0.0009 on MiniMax M2.5; repository review costs $0.195 and $0.0186; the cache-heavy agent loop costs $0.27 and $0.078. MiniMax M2.5 does not fit this workload in one request. MiniMax M2.5 has no published cached-input rate, so cached tokens use its listed input rate.

Which has the larger context window, Kimi K3 or MiniMax M2.5?

Kimi K3 has the larger documented context window: 1.05M, compared with 128K.

Self-host vs API cost

Estimates at 50,000 req/day · 1000 tokens/req average.

Kimi K3
API / mo$13,500
Self-host / mo$57,606
Break-evenNot reached at 70% load
MiniMax M2.5
API / mo$1,125
Self-host / moNot listed
Break-even—
Proprietary model — self-hosting not applicable.
Model the full break-even

Benchmark evidence

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

Browse raw public benchmark evidence61 rows

Agentic

  • Terminal-Bench 4.1

    Kimi K323.00%
    Source
    MiniMax M2.5—

    Not directly comparable

  • Finch

    Kimi K377.3%
    Source
    MiniMax M2.5—

    Not directly comparable

  • b3 Agent Security

    Kimi K388.0%
    Source
    MiniMax M2.5—

    Not directly comparable

  • Cyber Misuse Refusal

    Kimi K394.0%
    Source
    MiniMax M2.5—

    Not directly comparable

  • Terminal-Bench 2.1

    Kimi K388.3%
    Source
    MiniMax M2.5—

    Not directly comparable

  • BrowseComp

    Kimi K391.2%
    Source
    MiniMax M2.5—

    Not directly comparable

  • DeepSearchQA

    Kimi K395.0%
    Source
    MiniMax M2.5—

    Not directly comparable

  • Toolathlon-Verified

    Kimi K373.2%
    Source
    MiniMax M2.5—

    Not directly comparable

  • MCP Atlas

    Kimi K384.2%
    Source
    MiniMax M2.5—

    Not directly comparable

  • AutomationBench

    Kimi K330.8%
    Source
    MiniMax M2.5—

    Not directly comparable

  • JobBench

    Kimi K352.9%
    Source
    MiniMax M2.5—

    Not directly comparable

  • APEX-Agents

    Kimi K337.6%
    Source
    MiniMax M2.5—

    Not directly comparable

  • SpreadsheetBench 2

    Kimi K334.8%
    Source
    MiniMax M2.5—

    Not directly comparable

  • DECK-Bench

    Kimi K373.5%
    Source
    MiniMax M2.5—

    Not directly comparable

  • Terminal-Bench 2.1 (Vals)

    Kimi K380.9%
    Source
    MiniMax M2.5—

    Not directly comparable

  • ApprenticeBench

    Kimi K318%
    Source
    MiniMax M2.5—

    Not directly comparable

  • Hermes Index

    Kimi K339.2%
    Source
    MiniMax M2.5—

    Not directly comparable

  • Hermes Bench

    Kimi K371.5%
    Source
    MiniMax M2.5—

    Not directly comparable

  • MazeBench

    Kimi K31%
    Source
    MiniMax M2.5—

    Not directly comparable

Coding

  • ScarfBench

    Kimi K361.3%
    Source
    MiniMax M2.5—

    Not directly comparable

  • DeepSWE

    Kimi K367.5%
    Source
    MiniMax M2.5—

    Not directly comparable

  • CursorBench 3.2

    Kimi K360.8%
    Source
    MiniMax M2.5—

    Not directly comparable

  • FrontierSWE

    Kimi K381.2%
    Source
    MiniMax M2.5—

    Not directly comparable

  • ProgramBench

    Kimi K377.8%
    Source
    MiniMax M2.5—

    Not directly comparable

  • Kimi Code Bench v2

    Kimi K372.9%
    Source
    MiniMax M2.5—

    Not directly comparable

  • sweMarathon

    Kimi K342%
    Source
    MiniMax M2.5—

    Not directly comparable

  • PostTrainBench (lab runs)

    Kimi K336.6%
    Source
    MiniMax M2.5—

    Not directly comparable

  • MLS-Bench Lite

    Kimi K348.3%
    Source
    MiniMax M2.5—

    Not directly comparable

  • VulcanBench v3

    Kimi K373.7%
    Source
    MiniMax M2.5—

    Not directly comparable

  • OpenHarmony Bench v1.0

    Kimi K357.3%
    Source
    MiniMax M2.5—

    Not directly comparable

  • FrontierSWE v2

    Kimi K325.9%
    Source
    MiniMax M2.5—

    Not directly comparable

  • LiveCodeBench (Vals)

    Kimi K387.2%
    Source
    MiniMax M2.5—

    Not directly comparable

  • SWE-bench (Vals)

    Kimi K393.4%
    Source
    MiniMax M2.5—

    Not directly comparable

  • PostTrainBench v1.1

    Kimi K332.0%
    Source
    MiniMax M2.5—

    Not directly comparable

  • Bug Hunt Bench

    Kimi K321.0 fixes
    Source
    MiniMax M2.5—

    Not directly comparable

  • FrontierCode 2.0

    Kimi K318.9%
    Source
    MiniMax M2.5—

    Not directly comparable

  • PostTrainBench

    Kimi K331.5%
    Source
    MiniMax M2.5—

    Not directly comparable

  • Vibe Code Bench

    Kimi K3—
    MiniMax M2.514.85%
    Source

    Not directly comparable

Reasoning

  • ARC-AGI-1

    Kimi K394.50%
    Source
    MiniMax M2.5—

    Not directly comparable

  • ARC-AGI-2

    Kimi K360.4%
    Source
    MiniMax M2.5—

    Not directly comparable

Multimodal

  • ExtractBench (LlamaIndex)

    Kimi K383.17%
    Source
    MiniMax M2.5—

    Not directly comparable

  • OfficeQA Pro

    Kimi K363.3%
    Source
    MiniMax M2.5—

    Not directly comparable

  • MMMU-Pro

    Kimi K381.6%
    Source
    MiniMax M2.5—

    Not directly comparable

  • MMMU-Pro w/ Python

    Kimi K383.4%
    Source
    MiniMax M2.5—

    Not directly comparable

  • CharXiv w/o tools

    Kimi K384.8%
    Source
    MiniMax M2.5—

    Not directly comparable

  • CharXiv

    Kimi K391.3%
    Source
    MiniMax M2.5—

    Not directly comparable

  • MathVision

    Kimi K394.3%
    Source
    MiniMax M2.5—

    Not directly comparable

  • MathVision w/ Python

    Kimi K397.8%
    Source
    MiniMax M2.5—

    Not directly comparable

  • BabyVision w/ Python

    Kimi K385.7%
    Source
    MiniMax M2.5—

    Not directly comparable

  • ZeroBench

    Kimi K323.0%
    Source
    MiniMax M2.5—

    Not directly comparable

  • ZeroBench w/ Python

    Kimi K341.0%
    Source
    MiniMax M2.5—

    Not directly comparable

  • WorldVQA ForceAnswer

    Kimi K351.0%
    Source
    MiniMax M2.5—

    Not directly comparable

  • OmniDocBench

    Kimi K391.1%
    Source
    MiniMax M2.5—

    Not directly comparable

  • PerceptionBench

    Kimi K358.5%
    Source
    MiniMax M2.5—

    Not directly comparable

Knowledge

  • GPQA

    Kimi K393.5%
    Source
    MiniMax M2.5—

    Not directly comparable

  • GPQA-D

    Kimi K393.5%
    Source
    MiniMax M2.5—

    Not directly comparable

  • HLE

    Kimi K356%
    Source
    MiniMax M2.5—

    Not directly comparable

  • HLE w/o tools

    Kimi K343.5%
    Source
    MiniMax M2.5—

    Not directly comparable

  • GPQA Diamond (Vals)

    Kimi K392.9%
    Source
    MiniMax M2.5—

    Not directly comparable

  • MMLU-Pro (Vals)

    Kimi K388.0%
    Source
    MiniMax M2.5—

    Not directly comparable

Instruction following

  • Gray Swan IPI (15 attempts)

    Kimi K352.7%
    Source
    MiniMax M2.5—

    Not directly comparable

61 public results · 0 shared

Watch Kimi K3 vs MiniMax M2.5

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

Read a sample issue

Join 7,000+ subscribers.

Last updated November 17, 2026