Coding
Like-for-like- MiMo-V2.5
- 43.5
- Supported · #101/152
- Qwen3.8 Max
- 60.8
- Supported · #20/152
- Basis
- BenchAlign lane · 4 vs 12 public rows
- Reading
- Qwen3.8 Max leads · intervals overlap
MiMo-V2.5 vs Qwen3.8 Max
Decision reading
Qwen3.8 Max has the higher public score estimate, 71.76 versus 57.88, but the 90% score intervals overlap. Treat that as a lead, not a settled winner.
9 results are shared. Category rows resting on Estimated evidence or different benchmark sets are marked directional and do not name a winner.
Updated September 15, 2026. Rank says Qwen3.8 Max is ahead. Price, access, and your workload can each overturn that. Public scores include evidence status and uncertainty.
Share or export
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.
Coding work
Code generation, repair, and software-engineering tasks
Qwen3.8 Max
Qwen3.8 Max leads on the public coding lane, 60.8 to 43.5, with Supported evidence for both models, although the 90% intervals overlap.
Confidence: limited
Show secondary and unsupported calls
Agentic work
Tool use, computer use, and multi-step task completion
Not enough matched evidence
MiMo-V2.5 is scored on Estimated evidence for agentic, so the reading is directional rather than like-for-like.
Confidence: limited
Long documents
Prompts that approach the documented context limit
No clear pick
The documented context windows are equal.
Confidence: documented
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
What is actually comparable
Shared results can support a head-to-head reading. Results present for only one model describe coverage, not superiority.
6 MiMo-V2.5 only
9 Shared
51 Qwen3.8 Max only
- Shared results
- 9
- MiMo-V2.5 only
- 6
- Qwen3.8 Max only
- 51
- Like-for-like categories
- 2 / 8
2 categories rest on Estimated evidence or different evidence sets. Those rows remain visible for coverage context but do not name a winner.
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.
Multimodal
Like-for-like- MiMo-V2.5
- 59.2
- #29/48
- Qwen3.8 Max
- 87.4
- #5/48
- Basis
- Provisional lane · 2 vs 2 weighted rows
- Reading
- Qwen3.8 Max leads
Agentic
Directional only- MiMo-V2.5
- 49.6
- Estimated · #60/153
- Qwen3.8 Max
- 67.3
- Supported · #9/153
- Basis
- BenchAlign lane · 6 vs 15 public rows
- Reading
- Directional only
Knowledge
Directional only- MiMo-V2.5
- 52.2
- Estimated · #67/183
- Qwen3.8 Max
- 68.8
- Supported · #17/183
- Basis
- BenchAlign lane · 2 vs 6 public rows
- Reading
- Directional only
Reasoning
Not comparable- MiMo-V2.5
- Not ranked
- Qwen3.8 Max
- 86.5
- Unranked · 2 rankable rows
- Basis
- Provisional lane · 0 vs 2 weighted rows
- Reading
- Not comparable
Math
Not comparable- MiMo-V2.5
- Not ranked
- Qwen3.8 Max
- Not ranked
- Basis
- Provisional lane · 0 vs 0 weighted rows
- Reading
- Not comparable
Multilingual
Not comparable- MiMo-V2.5
- Not ranked
- Qwen3.8 Max
- Not ranked
- Basis
- Provisional lane · 0 vs 0 weighted rows
- Reading
- Not comparable
Instruction following
Not comparable- MiMo-V2.5
- Not ranked
- Qwen3.8 Max
- 90.7
- #18/123
- Basis
- Provisional lane · 0 vs 1 weighted rows
- Reading
- Not comparable
| Category | MiMo-V2.5 | Qwen3.8 Max | Basis | Reading |
|---|---|---|---|---|
| Coding | 43.5Supported · #101/152 | 60.8Supported · #20/152 | Like-for-likeBenchAlign lane · 4 vs 12 public rows | Qwen3.8 Max leads · intervals overlap |
| Multimodal | 59.2#29/48 | 87.4#5/48 | Like-for-likeProvisional lane · 2 vs 2 weighted rows | Qwen3.8 Max leads |
| Agentic | 49.6Estimated · #60/153 | 67.3Supported · #9/153 | Directional onlyBenchAlign lane · 6 vs 15 public rows | Directional only |
| Knowledge | 52.2Estimated · #67/183 | 68.8Supported · #17/183 | Directional onlyBenchAlign lane · 2 vs 6 public rows | Directional only |
| Reasoning | Not ranked | 86.5Unranked · 2 rankable rows | Not comparableProvisional lane · 0 vs 2 weighted rows | Not comparable |
| Math | Not ranked | Not ranked | Not comparableProvisional lane · 0 vs 0 weighted rows | Not comparable |
| Multilingual | Not ranked | Not ranked | Not comparableProvisional lane · 0 vs 0 weighted rows | Not comparable |
| Instruction following | Not ranked | 90.7#18/123 | Not comparableProvisional lane · 0 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.
CharXiv
Multimodal
SWE-bench Pro
Coding
LiveCodeBench (Vals)
Coding
MMLU-Pro (Vals)
Knowledge
MMMU-Pro
Multimodal
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
- MiMo-V2.5
- API rate not published
- Fits in one request
- Qwen3.8 Max
- API rate not published
- Fits in one request
MiMo-V2.5 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
- MiMo-V2.5
- API rate not published
- Fits in one request
- Qwen3.8 Max
- API rate not published
- Fits in one request
MiMo-V2.5 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
- MiMo-V2.5
- API rate not published
- Fits in one request
- Cached-input rate unavailable
- Qwen3.8 Max
- API rate not published
- Fits in one request
- Cached-input rate unavailable
MiMo-V2.5 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.
SpecificationMiMo-V2.5Qwen3.8 Max
Context window
Maximum documented context; output-token limits may be lower.
MiMo-V2.5
1M
Qwen3.8 Max
API model ID
MiMo-V2.5
Not sourced
Qwen3.8 Max
qwen3.8-max
Alibaba Cloud Model Studio pricingCached-input rate
A missing cached-input rate falls back to the listed input rate only in the stated workload estimate.
MiMo-V2.5
No comparable hosted API rate
Qwen3.8 Max
No comparable hosted API rate
Alibaba Cloud Model Studio pricingDocumented inputs
MiMo-V2.5
Not sourced
Qwen3.8 Max
Not sourced
Documented outputs
MiMo-V2.5
Not sourced
Qwen3.8 Max
Not sourced
Provider availability
MiMo-V2.5
Not sourced
Qwen3.8 Max
Not sourced
Reasoning profile
MiMo-V2.5
Reasoning
Qwen3.8 Max
Reasoning
Weight access
MiMo-V2.5
Proprietary
Qwen3.8 Max
Open Weight
License
MiMo-V2.5
Proprietary
Qwen3.8 Max
Open Weight
Release date
MiMo-V2.5
2026-04-22
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
- Qwen3.8 Max has the higher public score estimate, 71.76 versus 57.88, but the 90% score intervals overlap.
- Workload cost
- A complete comparable API-rate estimate is not available for both models.
- Context tradeoff
- Both models list 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 evidence66 rows
Agentic
Claw-Eval
MiMo-V2.562.3%SourceQwen3.8 Max—Not directly comparable
MM-ClawBench
MiMo-V2.523.8%SourceQwen3.8 Max—Not directly comparable
Terminal-Bench 2.0
MiMo-V2.565.8%SourceQwen3.8 Max—Not directly comparable
Gert Labs
MiMo-V2.546.89%SourceQwen3.8 Max—Not directly comparable
ResearchClawBench
MiMo-V2.516.9%SourceQwen3.8 Max—Not directly comparable
Terminal-Bench 2.1 (Vals)
MiMo-V2.560.7%SourceQwen3.8 Max67.4%SourceQwen3.8 Max leads this result
Terminal-Bench 2.1
MiMo-V2.5—Qwen3.8 Max86.6%SourceNot directly comparable
CoWorkBench
MiMo-V2.5—Qwen3.8 Max74.8%SourceNot directly comparable
JobBench
MiMo-V2.5—Qwen3.8 Max53.4%SourceNot directly comparable
skillsBench
MiMo-V2.5—Qwen3.8 Max70.2%SourceNot directly comparable
Agents' Last Exam
MiMo-V2.5—Qwen3.8 Max52.4%SourceNot directly comparable
AutomationBench
MiMo-V2.5—Qwen3.8 Max27.3%SourceNot directly comparable
Toolathlon-Verified
MiMo-V2.5—Qwen3.8 Max72.5%SourceNot directly comparable
WideResearch
MiMo-V2.5—Qwen3.8 Max81.9%SourceNot directly comparable
HLE w/ tools
MiMo-V2.5—Qwen3.8 Max56.2%SourceNot directly comparable
OSWorld-Verified
MiMo-V2.5—Qwen3.8 Max86.1%SourceNot directly comparable
OSWorld 2.0
MiMo-V2.5—Qwen3.8 Max19.4%SourceNot directly comparable
WebArena-Verified
MiMo-V2.5—Qwen3.8 Max66.8%SourceNot directly comparable
AndroidWorld
MiMo-V2.5—Qwen3.8 Max85.3%SourceNot directly comparable
MobileWorld
MiMo-V2.5—Qwen3.8 Max77.8%SourceNot directly comparable
Coding
SWE-bench Pro
MiMo-V2.556.1%SourceQwen3.8 Max67.7%SourceQwen3.8 Max leads this result
Terminal-Bench 2.0
MiMo-V2.565.8%SourceQwen3.8 Max—Not directly comparable
LiveCodeBench (Vals)
MiMo-V2.581.5%SourceQwen3.8 Max87.9%SourceQwen3.8 Max leads this result
SWE-bench (Vals)
MiMo-V2.571.0%SourceQwen3.8 Max85.6%SourceQwen3.8 Max leads this result
Terminal-Bench 2.1
MiMo-V2.5—Qwen3.8 Max86.6%SourceNot directly comparable
DeepSWE
MiMo-V2.5—Qwen3.8 Max56.6%SourceNot directly comparable
NL2Repo
MiMo-V2.5—Qwen3.8 Max55.9%SourceNot directly comparable
FrontierSWE
MiMo-V2.5—Qwen3.8 Max73.5%SourceNot directly comparable
MLS-Bench Lite
MiMo-V2.5—Qwen3.8 Max41.0%SourceNot directly comparable
PaperBench
MiMo-V2.5—Qwen3.8 Max93.0%SourceNot directly comparable
VulcanBench v3
MiMo-V2.5—Qwen3.8 Max81.2%SourceNot directly comparable
OpenHarmony Bench
MiMo-V2.5—Qwen3.8 Max60.8%SourceNot directly comparable
FrontierSWE v2
MiMo-V2.5—Qwen3.8 Max15.8%SourceNot directly comparable
Reasoning
Knowledge
GPQA Diamond (Vals)
MiMo-V2.581.6%SourceQwen3.8 Max93.7%SourceQwen3.8 Max leads this result
MMLU-Pro (Vals)
MiMo-V2.582.9%SourceQwen3.8 Max88.6%SourceQwen3.8 Max leads this result
GPQA
MiMo-V2.5—Qwen3.8 Max92.6%SourceNot directly comparable
GPQA-D
MiMo-V2.5—Qwen3.8 Max92.6%SourceNot directly comparable
HLE
MiMo-V2.5—Qwen3.8 Max43.6%SourceNot directly comparable
HLE w/o tools
MiMo-V2.5—Qwen3.8 Max43.6%SourceNot directly comparable
Multimodal
Video-MME (with subtitle)
MiMo-V2.587.7%SourceQwen3.8 Max90.4%SourceQwen3.8 Max leads this result
CharXiv
MiMo-V2.581%SourceQwen3.8 Max93.5%SourceQwen3.8 Max leads this result
MMMU-Pro
MiMo-V2.577.9%SourceQwen3.8 Max82.3%SourceQwen3.8 Max leads this result
MathVision
MiMo-V2.5—Qwen3.8 Max95.2%SourceNot directly comparable
MathVision w/ Python
MiMo-V2.5—Qwen3.8 Max97.7%SourceNot directly comparable
BabyVision
MiMo-V2.5—Qwen3.8 Max82.0%SourceNot directly comparable
BabyVision w/ Python
MiMo-V2.5—Qwen3.8 Max91.3%SourceNot directly comparable
ZeroBench
MiMo-V2.5—Qwen3.8 Max24.0%SourceNot directly comparable
ZeroBench w/ Python
MiMo-V2.5—Qwen3.8 Max49.0%SourceNot directly comparable
MedXpertQA (MM)
MiMo-V2.5—Qwen3.8 Max80.4%SourceNot directly comparable
ScreenSpot Pro
MiMo-V2.5—Qwen3.8 Max84.5%SourceNot directly comparable
Vision2Web
MiMo-V2.5—Qwen3.8 Max69.0%SourceNot directly comparable
CharXiv w/o tools
MiMo-V2.5—Qwen3.8 Max88.4%SourceNot directly comparable
OmniDocBench 1.5
MiMo-V2.5—Qwen3.8 Max92.1%SourceNot directly comparable
OCRBench V2
MiMo-V2.5—Qwen3.8 Max74.2%SourceNot directly comparable
CC-OCR
MiMo-V2.5—Qwen3.8 Max79.6%SourceNot directly comparable
RealWorldQA
MiMo-V2.5—Qwen3.8 Max88.0%SourceNot directly comparable
ERQA
MiMo-V2.5—Qwen3.8 Max77.8%SourceNot directly comparable
SimpleVQA
MiMo-V2.5—Qwen3.8 Max75.0%SourceNot directly comparable
PerceptionBench
MiMo-V2.5—Qwen3.8 Max63.5%SourceNot directly comparable
VideoMMMU
MiMo-V2.5—Qwen3.8 Max88.7%SourceNot directly comparable
MMVU
MiMo-V2.5—Qwen3.8 Max82.4%SourceNot directly comparable
MLVU (M-Avg)
MiMo-V2.5—Qwen3.8 Max90.8%SourceNot directly comparable
LVBench
MiMo-V2.5—Qwen3.8 Max81.8%SourceNot directly comparable
Instruction following
IFBench
MiMo-V2.5—Qwen3.8 Max82.8%SourceNot directly comparable
Questions
Which is better, MiMo-V2.5 or Qwen3.8 Max?
Qwen3.8 Max has the higher public score estimate, 71.76 versus 57.88, but the 90% score intervals overlap. The higher estimate is not a decisive winner because the uncertainty ranges overlap.
Which is better for coding, MiMo-V2.5 or Qwen3.8 Max?
Qwen3.8 Max leads the public coding lane, 60.8 to 43.5, with Supported evidence for both models, although the 90% intervals overlap.
Which is better for agentic tasks, MiMo-V2.5 or Qwen3.8 Max?
Qwen3.8 Max scores higher for agentic tasks on the public lane, 67.3 to 49.6. MiMo-V2.5 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, MiMo-V2.5 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, MiMo-V2.5 or Qwen3.8 Max?
Both models list the same context window, 1M.
Related comparisons
Last updated September 15, 2026
Watch MiMo-V2.5 vs Qwen3.8 Max
One weekly email when material rank, price, or benchmark evidence changes make this matchup worth revisiting.
Read a sample issueJoin 2,000+ readers.