Model A

OpenBMB

—

Evidence status unavailable

90% interval unavailable

MiniCPM5-2B vs Qwen3.8 Max

Updated September 10, 2026. Public scores include evidence status and uncertainty. They are not guarantees for a specific workload.

Model B

Qwen3.8 Max

Alibaba

71.63/100

Supported · Public rank #11

90% interval 68.0–75.3

Decision reading

At least one model is not scored in the current public ranking lane, so the page does not name an overall quality winner.

7 results are shared. Category rows resting on Estimated evidence or different benchmark sets are marked directional and do not name a winner.

Share or export

Share on X LinkedInSocial card CSV JSON

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
Qwen3.8 Max
Qwen3.8 Max 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
MiniCPM5-2B is not ranked on the public lane for coding, so no winner is named for coding.
Confidence: limited
Agentic work
Tool use, computer use, and multi-step task completion
Not enough matched evidence
MiniCPM5-2B 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
The page does not recommend a cost winner because at least one model cannot fit the stated workload in one request. MiniCPM5-2B does not fit this workload in one request. MiniCPM5-2B has no comparable published API token rate. Qwen3.8 Max has no comparable published API token rate.
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

What is actually comparable

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

12 MiniCPM5-2B only

7 Shared

53 Qwen3.8 Max only

Shared results: 7
MiniCPM5-2B only: 12
Qwen3.8 Max only: 53
Like-for-like categories: 0 / 8

Category results, on a stated basis

Each row shows the public-lane category score for both models: the BenchAlign 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.

Agentic

Not comparable

MiniCPM5-2B: Not ranked
Qwen3.8 Max: 67.2; Supported · #9/152
Basis: BenchAlign lane · 2 vs 15 public rows
Reading: Not comparable

Coding

Not comparable

MiniCPM5-2B: Not ranked
Qwen3.8 Max: 60.8; Supported · #20/151
Basis: BenchAlign lane · 5 vs 12 public rows
Reading: Not comparable

Reasoning

Not comparable

MiniCPM5-2B: Not ranked
Qwen3.8 Max: 86.5; Unranked · 2 rankable rows
Basis: Provisional lane · 1 vs 2 weighted rows
Reading: Not comparable

Knowledge

Not comparable

MiniCPM5-2B: Not ranked
Qwen3.8 Max: 68.8; Supported · #17/183
Basis: BenchAlign lane · 5 vs 6 public rows
Reading: Not comparable

Math

Not comparable

MiniCPM5-2B: Not ranked
Qwen3.8 Max: Not ranked
Basis: Provisional lane · 2 vs 0 weighted rows
Reading: Not comparable

Multilingual

Not comparable

MiniCPM5-2B: Not ranked
Qwen3.8 Max: Not ranked
Basis: Provisional lane · 0 vs 0 weighted rows
Reading: Not comparable

Multimodal

Not comparable

MiniCPM5-2B: Not ranked
Qwen3.8 Max: 87.4; #5/48
Basis: Provisional lane · 0 vs 2 weighted rows
Reading: Not comparable

Instruction following

Not comparable

MiniCPM5-2B: Not ranked
Qwen3.8 Max: 90.7; #18/123
Basis: Provisional lane · 1 vs 1 weighted rows
Reading: Not comparable

Category scores with the server-provided evidence basis for MiniCPM5-2B and Qwen3.8 Max
Category	MiniCPM5-2B	Qwen3.8 Max	Basis	Reading
Agentic	Not ranked	67.2Supported · #9/152	Not comparableBenchAlign lane · 2 vs 15 public rows	Not comparable
Coding	Not ranked	60.8Supported · #20/151	Not comparableBenchAlign lane · 5 vs 12 public rows	Not comparable
Reasoning	Not ranked	86.5Unranked · 2 rankable rows	Not comparableProvisional lane · 1 vs 2 weighted rows	Not comparable
Knowledge	Not ranked	68.8Supported · #17/183	Not comparableBenchAlign lane · 5 vs 6 public rows	Not comparable
Math	Not ranked	Not ranked	Not comparableProvisional lane · 2 vs 0 weighted rows	Not comparable
Multilingual	Not ranked	Not ranked	Not comparableProvisional lane · 0 vs 0 weighted rows	Not comparable
Multimodal	Not ranked	87.4#5/48	Not comparableProvisional lane · 0 vs 2 weighted rows	Not comparable
Instruction following	Not ranked	90.7#18/123	Not comparableProvisional lane · 1 vs 1 weighted rows	Not comparable

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

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.

SWE-bench Pro
Coding
MiniCPM5-2B: 14.4%Qwen3.8 Max: 67.7%Normalized gap 53.3MiniCPM5-2B source Qwen3.8 Max source
HLE
Knowledge
MiniCPM5-2B: 8.9%Qwen3.8 Max: 43.6%Normalized gap 34.7MiniCPM5-2B source Qwen3.8 Max source
LongBench v2
Reasoning
MiniCPM5-2B: 43.7%Qwen3.8 Max: 66.3%Normalized gap 22.6MiniCPM5-2B source Qwen3.8 Max source
IFBench
Instruction following
MiniCPM5-2B: 66.3%Qwen3.8 Max: 82.8%Normalized gap 16.5MiniCPM5-2B source Qwen3.8 Max source

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

MiniCPM5-2B: Self-hosted; infrastructure cost varies; Fits in one request
Qwen3.8 Max: API rate not published; Fits in one request

MiniCPM5-2B has no comparable published API token rate. Qwen3.8 Max has no comparable published API token rate.

Repository review

50K fresh input + 3K output tokens

MiniCPM5-2B: Self-hosted; infrastructure cost varies; Fits in one request
Qwen3.8 Max: API rate not published; Fits in one request

MiniCPM5-2B has no comparable published API token rate. Qwen3.8 Max has no comparable published API token rate.

Cache-heavy agent loop

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

MiniCPM5-2B: Self-hosted; infrastructure cost varies; Does not fit in one request; Cached-input rate unavailable
Qwen3.8 Max: API rate not published; Fits in one request; Cached-input rate unavailable

MiniCPM5-2B does not fit this workload in one request. MiniCPM5-2B has no comparable published API token rate. Qwen3.8 Max has no comparable published API token rate.

Specification differences

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

SpecificationMiniCPM5-2BQwen3.8 Max

Context window

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

MiniCPM5-2B

131K

OpenBMB MiniCPM5-2B model card

Qwen3.8 Max

Alibaba Cloud Model Studio pricing

API model ID

MiniCPM5-2B

openbmb/MiniCPM5-2B

OpenBMB MiniCPM5-2B model card

Qwen3.8 Max

qwen3.8-max

Alibaba Cloud Model Studio pricing

Cached-input rate

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

MiniCPM5-2B

No comparable hosted API rate

OpenBMB MiniCPM5-2B model card

Qwen3.8 Max

No comparable hosted API rate

Alibaba Cloud Model Studio pricing

Documented inputs

MiniCPM5-2B

Not sourced

Qwen3.8 Max

Not sourced

Documented outputs

MiniCPM5-2B

Not sourced

Qwen3.8 Max

Not sourced

Provider availability

MiniCPM5-2B

Not sourced

Qwen3.8 Max

Not sourced

Reasoning profile

MiniCPM5-2B

Reasoning

Qwen3.8 Max

Reasoning

Weight access

MiniCPM5-2B

Open Weight

Qwen3.8 Max

Open Weight

License

MiniCPM5-2B

Open Weight

Qwen3.8 Max

Open Weight

Release date

MiniCPM5-2B

2026-09-06

Qwen3.8 Max

2026-08-03

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: At least one model is not scored in the current public ranking lane, so the page does not name an overall quality winner.
Workload cost: A complete comparable API-rate estimate is not available for both models.
Context tradeoff: Qwen3.8 Max has the larger documented window (1M).

Run the same representative tasks against both endpoints before changing production traffic.

Benchmark evidence

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

Browse raw public benchmark evidence72 rows

Agentic

BFCL v4
MiniCPM5-2B66.6%
Source
Qwen3.8 Max—
Not directly comparable
Terminal-Bench 2.1
MiniCPM5-2B8.6%
Source
Qwen3.8 Max86.6%
Source
Qwen3.8 Max leads this result
CoWorkBench
MiniCPM5-2B—
Qwen3.8 Max74.8%
Source
Not directly comparable
JobBench
MiniCPM5-2B—
Qwen3.8 Max53.4%
Source
Not directly comparable
skillsBench
MiniCPM5-2B—
Qwen3.8 Max70.2%
Source
Not directly comparable
Agents' Last Exam
MiniCPM5-2B—
Qwen3.8 Max52.4%
Source
Not directly comparable
AutomationBench
MiniCPM5-2B—
Qwen3.8 Max27.3%
Source
Not directly comparable
Toolathlon-Verified
MiniCPM5-2B—
Qwen3.8 Max72.5%
Source
Not directly comparable
WideResearch
MiniCPM5-2B—
Qwen3.8 Max81.9%
Source
Not directly comparable
HLE w/ tools
MiniCPM5-2B—
Qwen3.8 Max56.2%
Source
Not directly comparable
OSWorld-Verified
MiniCPM5-2B—
Qwen3.8 Max86.1%
Source
Not directly comparable
OSWorld 2.0
MiniCPM5-2B—
Qwen3.8 Max19.4%
Source
Not directly comparable
WebArena-Verified
MiniCPM5-2B—
Qwen3.8 Max66.8%
Source
Not directly comparable
AndroidWorld
MiniCPM5-2B—
Qwen3.8 Max85.3%
Source
Not directly comparable
MobileWorld
MiniCPM5-2B—
Qwen3.8 Max77.8%
Source
Not directly comparable
Terminal-Bench 2.1 (Vals)
MiniCPM5-2B—
Qwen3.8 Max67.4%
Source
Not directly comparable

Coding

LiveCodeBench v6
MiniCPM5-2B69.1%
Source
Qwen3.8 Max—
Not directly comparable
SciCode
MiniCPM5-2B26.3%
Source
Qwen3.8 Max—
Not directly comparable
SWE-bench Verified
MiniCPM5-2B46.4%
Source
Qwen3.8 Max—
Not directly comparable
SWE-bench Pro
MiniCPM5-2B14.4%
Source
Qwen3.8 Max67.7%
Source
Qwen3.8 Max leads this result
Terminal-Bench 2.1
MiniCPM5-2B8.6%
Source
Qwen3.8 Max86.6%
Source
Qwen3.8 Max leads this result
DeepSWE
MiniCPM5-2B—
Qwen3.8 Max56.6%
Source
Not directly comparable
NL2Repo
MiniCPM5-2B—
Qwen3.8 Max55.9%
Source
Not directly comparable
FrontierSWE
MiniCPM5-2B—
Qwen3.8 Max73.5%
Source
Not directly comparable
MLS-Bench Lite
MiniCPM5-2B—
Qwen3.8 Max41.0%
Source
Not directly comparable
PaperBench
MiniCPM5-2B—
Qwen3.8 Max93.0%
Source
Not directly comparable
VulcanBench v3
MiniCPM5-2B—
Qwen3.8 Max81.2%
Source
Not directly comparable
OpenHarmony Bench
MiniCPM5-2B—
Qwen3.8 Max60.8%
Source
Not directly comparable
FrontierSWE v2
MiniCPM5-2B—
Qwen3.8 Max15.8%
Source
Not directly comparable
LiveCodeBench (Vals)
MiniCPM5-2B—
Qwen3.8 Max87.9%
Source
Not directly comparable
SWE-bench (Vals)
MiniCPM5-2B—
Qwen3.8 Max85.6%
Source
Not directly comparable

Reasoning

LongBench v2
MiniCPM5-2B43.7%
Source
Qwen3.8 Max66.3%
Source
Qwen3.8 Max leads this result
MRCRv2
MiniCPM5-2B—
Qwen3.8 Max92.9%
Source
Not directly comparable

Knowledge

MMLU-Pro
MiniCPM5-2B70.8%
Source
Qwen3.8 Max—
Not directly comparable
MMLU-Redux
MiniCPM5-2B84.7%
Source
Qwen3.8 Max—
Not directly comparable
HLE
MiniCPM5-2B8.9%
Source
Qwen3.8 Max43.6%
Source
Qwen3.8 Max leads this result
GPQA-D
MiniCPM5-2B70.2%
Source
Qwen3.8 Max92.6%
Source
Qwen3.8 Max leads this result
SuperGPQA
MiniCPM5-2B40.8%
Source
Qwen3.8 Max—
Not directly comparable
GPQA
MiniCPM5-2B—
Qwen3.8 Max92.6%
Source
Not directly comparable
HLE w/o tools
MiniCPM5-2B—
Qwen3.8 Max43.6%
Source
Not directly comparable
GPQA Diamond (Vals)
MiniCPM5-2B—
Qwen3.8 Max93.7%
Source
Not directly comparable
MMLU-Pro (Vals)
MiniCPM5-2B—
Qwen3.8 Max88.6%
Source
Not directly comparable

Math

AIME 2025
MiniCPM5-2B86.5%
Source
Qwen3.8 Max—
Not directly comparable
AIME26
MiniCPM5-2B86.5%
Source
Qwen3.8 Max—
Not directly comparable
HMMT Feb 2026
MiniCPM5-2B63.8%
Source
Qwen3.8 Max—
Not directly comparable
MATH-500
MiniCPM5-2B94.6%
Source
Qwen3.8 Max—
Not directly comparable

Multimodal

MMMU-Pro
MiniCPM5-2B—
Qwen3.8 Max82.3%
Source
Not directly comparable
MathVision
MiniCPM5-2B—
Qwen3.8 Max95.2%
Source
Not directly comparable
MathVision w/ Python
MiniCPM5-2B—
Qwen3.8 Max97.7%
Source
Not directly comparable
BabyVision
MiniCPM5-2B—
Qwen3.8 Max82.0%
Source
Not directly comparable
BabyVision w/ Python
MiniCPM5-2B—
Qwen3.8 Max91.3%
Source
Not directly comparable
ZeroBench
MiniCPM5-2B—
Qwen3.8 Max24.0%
Source
Not directly comparable
ZeroBench w/ Python
MiniCPM5-2B—
Qwen3.8 Max49.0%
Source
Not directly comparable
MedXpertQA (MM)
MiniCPM5-2B—
Qwen3.8 Max80.4%
Source
Not directly comparable
ScreenSpot Pro
MiniCPM5-2B—
Qwen3.8 Max84.5%
Source
Not directly comparable
Vision2Web
MiniCPM5-2B—
Qwen3.8 Max69.0%
Source
Not directly comparable
CharXiv w/o tools
MiniCPM5-2B—
Qwen3.8 Max88.4%
Source
Not directly comparable
CharXiv
MiniCPM5-2B—
Qwen3.8 Max93.5%
Source
Not directly comparable
OmniDocBench 1.5
MiniCPM5-2B—
Qwen3.8 Max92.1%
Source
Not directly comparable
OCRBench V2
MiniCPM5-2B—
Qwen3.8 Max74.2%
Source
Not directly comparable
CC-OCR
MiniCPM5-2B—
Qwen3.8 Max79.6%
Source
Not directly comparable
RealWorldQA
MiniCPM5-2B—
Qwen3.8 Max88.0%
Source
Not directly comparable
ERQA
MiniCPM5-2B—
Qwen3.8 Max77.8%
Source
Not directly comparable
SimpleVQA
MiniCPM5-2B—
Qwen3.8 Max75.0%
Source
Not directly comparable
PerceptionBench
MiniCPM5-2B—
Qwen3.8 Max63.5%
Source
Not directly comparable
Video-MME (with subtitle)
MiniCPM5-2B—
Qwen3.8 Max90.4%
Source
Not directly comparable
VideoMMMU
MiniCPM5-2B—
Qwen3.8 Max88.7%
Source
Not directly comparable
MMVU
MiniCPM5-2B—
Qwen3.8 Max82.4%
Source
Not directly comparable
MLVU (M-Avg)
MiniCPM5-2B—
Qwen3.8 Max90.8%
Source
Not directly comparable
LVBench
MiniCPM5-2B—
Qwen3.8 Max81.8%
Source
Not directly comparable

Instruction following

IFBench
MiniCPM5-2B66.3%
Source
Qwen3.8 Max82.8%
Source
Qwen3.8 Max leads this result
IFEval
MiniCPM5-2B86.7%
Source
Qwen3.8 Max—
Not directly comparable

Frequently asked questions

Which is better, MiniCPM5-2B or Qwen3.8 Max?

At least one model is not scored in the current public ranking lane, so the page does not name an overall quality winner. The page therefore keeps the decision tied to the specific documented workload.

Which is better for coding, MiniCPM5-2B or Qwen3.8 Max?

MiniCPM5-2B is not ranked on the public lane for coding, so no winner is named for coding.

Which is better for agentic tasks, MiniCPM5-2B or Qwen3.8 Max?

MiniCPM5-2B is not ranked on the public lane for agentic tasks, so no winner is named for agentic tasks.

Which costs less, MiniCPM5-2B or Qwen3.8 Max?

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, MiniCPM5-2B or Qwen3.8 Max?

Qwen3.8 Max has the larger documented context window: 1M, compared with 131K.

Related comparisons

Compare API pricing Read the methodology Open the model selector

Last updated September 10, 2026

Watch MiniCPM5-2B vs Qwen3.8 Max

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

Read a sample issue

Join 2,000+ readers.

MiniCPM5-2B vs Qwen3.8 Max

Which one for your work

Long documents

Coding work

Agentic work

Chat turn cost

Cache-heavy agent loop cost

Repository review cost

What is actually comparable

Category results, on a stated basis

Agentic

Coding

Reasoning

Knowledge

Math

Multilingual

Multimodal

Instruction following

Shape of the matched evidence

What each workload costs

Chat turn

Repository review

Cache-heavy agent loop

Specification differences

Context window

API model ID

Cached-input rate

Documented inputs

Documented outputs

Provider availability

Reasoning profile

Weight access

License

Release date

Benchmark evidence

Agentic

Coding

Reasoning

Knowledge

Math

Multimodal

Instruction following

Frequently asked questions

Which is better, MiniCPM5-2B or Qwen3.8 Max?

Which is better for coding, MiniCPM5-2B or Qwen3.8 Max?

Which is better for agentic tasks, MiniCPM5-2B or Qwen3.8 Max?

Which costs less, MiniCPM5-2B or Qwen3.8 Max?

Which has the larger context window, MiniCPM5-2B or Qwen3.8 Max?

Related comparisons

Watch MiniCPM5-2B vs Qwen3.8 Max