Knowledge
Directional only- Kimi K2.5 (Reasoning)
- 87.2
- Qwen3.8 Max
- 50.2
- Weighted basis
- 2 vs 2 rows
- Reading
- Directional only
Model comparison
Kimi K2.5 (Reasoning) vs Qwen3.8 Max
Updated August 3, 2026. Public scores include evidence status and uncertainty. They are not guarantees for a specific workload.
Qwen3.8 Max has the higher public score estimate, 65.4 versus 58.61, but the 90% score intervals overlap. Treat that as a lead, not a settled winner.
2 results are shared. Category rows based on different benchmark sets are marked directional and do not name a winner.
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
No shared weighted benchmark basis supports a winner.
Confidence: limited
Agentic work
Tool use, computer use, and multi-step task completion
Not enough matched evidence
No shared weighted benchmark basis supports a winner.
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: rate-fallback
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.
7 Kimi K2.5 (Reasoning) only
2 Shared
50 Qwen3.8 Max only
- Shared results
- 2
- Kimi K2.5 (Reasoning) only
- 7
- Qwen3.8 Max only
- 50
- Like-for-like categories
- 0 / 8
2 categories use different evidence sets. Those rows remain visible for coverage context but do not name a winner.
Category results, on a stated basis
Each row states whether both averages use the same weighted benchmark set. Directional and not-comparable rows remain visible, but they never receive a winner in this template.
Multimodal
Directional only- Kimi K2.5 (Reasoning)
- 78.5
- Qwen3.8 Max
- 86.3
- Weighted basis
- 1 vs 2 rows
- Reading
- Directional only
Agentic
Not comparable- Kimi K2.5 (Reasoning)
- 55.0
- Qwen3.8 Max
- 86.1
- Weighted basis
- 2 vs 1 rows
- Reading
- Not comparable
Coding
Not comparable- Kimi K2.5 (Reasoning)
- 76.8
- Qwen3.8 Max
- 67.7
- Weighted basis
- 1 vs 1 rows
- Reading
- Not comparable
Reasoning
Not comparable- Kimi K2.5 (Reasoning)
- Not measured
- Qwen3.8 Max
- 78.3
- Weighted basis
- 0 vs 2 rows
- Reading
- Not comparable
Math
Not comparable- Kimi K2.5 (Reasoning)
- Not measured
- Qwen3.8 Max
- Not measured
- Weighted basis
- 0 vs 0 rows
- Reading
- Not comparable
Multilingual
Not comparable- Kimi K2.5 (Reasoning)
- Not measured
- Qwen3.8 Max
- Not measured
- Weighted basis
- 0 vs 0 rows
- Reading
- Not comparable
Instruction following
Not comparable- Kimi K2.5 (Reasoning)
- Not measured
- Qwen3.8 Max
- 82.8
- Weighted basis
- 0 vs 1 rows
- Reading
- Not comparable
| Category | Kimi K2.5 (Reasoning) | Qwen3.8 Max | Weighted basis | Reading |
|---|---|---|---|---|
| Knowledge | 87.2 | 50.2 | Directional only2 vs 2 rows | Directional only |
| Multimodal | 78.5 | 86.3 | Directional only1 vs 2 rows | Directional only |
| Agentic | 55.0 | 86.1 | Not comparable2 vs 1 rows | Not comparable |
| Coding | 76.8 | 67.7 | Not comparable1 vs 1 rows | Not comparable |
| Reasoning | Not measured | 78.3 | Not comparable0 vs 2 rows | Not comparable |
| Math | Not measured | Not measured | Not comparable0 vs 0 rows | Not comparable |
| Multilingual | Not measured | Not measured | Not comparable0 vs 0 rows | Not comparable |
| Instruction following | Not measured | 82.8 | Not comparable0 vs 1 rows | Not comparable |
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.
GPQA
Knowledge
Kimi K2.5 (Reasoning): 87.6%Qwen3.8 Max: 92.6%Normalized gap 5.0Kimi K2.5 (Reasoning) source Qwen3.8 Max sourceMMMU-Pro
Multimodal
Kimi K2.5 (Reasoning): 78.5%Qwen3.8 Max: 82.3%Normalized gap 3.8Kimi K2.5 (Reasoning) 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
- Kimi K2.5 (Reasoning)
- $0.0021
- Fits in one request
- Qwen3.8 Max
- API rate not published
- Fits in one request
Qwen3.8 Max has no comparable published API token rate.
Repository review
50K fresh input + 3K output tokens
- Kimi K2.5 (Reasoning)
- $0.039
- Fits in one request
- Qwen3.8 Max
- API rate not published
- Fits in one request
Qwen3.8 Max has no comparable published API token rate.
Cache-heavy agent loop
200K cached + 20K fresh input + 10K output tokens
- Kimi K2.5 (Reasoning)
- $0.162
- Fits in one request
- Cached input priced at the published list-input rate
- Qwen3.8 Max
- API rate not published
- Fits in one request
- Cached-input rate unavailable
Kimi K2.5 (Reasoning) has no published cached-input rate, so cached tokens use its listed input 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.
SpecificationKimi K2.5 (Reasoning)Qwen3.8 Max
Context window
Maximum documented context; output-token limits may be lower.
Kimi K2.5 (Reasoning)
256K
Qwen3.8 Max
API model ID
Kimi K2.5 (Reasoning)
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.
Kimi K2.5 (Reasoning)
Not published
Qwen3.8 Max
No comparable hosted API rate
Alibaba Cloud Model Studio pricingDocumented inputs
Kimi K2.5 (Reasoning)
Not sourced
Qwen3.8 Max
Not sourced
Documented outputs
Kimi K2.5 (Reasoning)
Not sourced
Qwen3.8 Max
Not sourced
Provider availability
Kimi K2.5 (Reasoning)
Not sourced
Qwen3.8 Max
Not sourced
Reasoning profile
Kimi K2.5 (Reasoning)
Reasoning
Qwen3.8 Max
Reasoning
Weight access
Kimi K2.5 (Reasoning)
Proprietary
Qwen3.8 Max
Proprietary
License
Kimi K2.5 (Reasoning)
Proprietary
Qwen3.8 Max
Proprietary
Release date
Kimi K2.5 (Reasoning)
2026-02-01
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, 65.4 versus 58.61, but the 90% score intervals overlap.
- 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 evidence59 rows
Agentic
Terminal-Bench 2.0
Kimi K2.5 (Reasoning)50.8%SourceQwen3.8 Max—Not directly comparable
BrowseComp
Kimi K2.5 (Reasoning)60.6%SourceQwen3.8 Max—Not directly comparable
Gert Labs
Kimi K2.5 (Reasoning)32.58%SourceQwen3.8 Max—Not directly comparable
Terminal-Bench 2.1
Kimi K2.5 (Reasoning)—Qwen3.8 Max86.6%SourceNot directly comparable
CoWorkBench
Kimi K2.5 (Reasoning)—Qwen3.8 Max74.8%SourceNot directly comparable
JobBench
Kimi K2.5 (Reasoning)—Qwen3.8 Max53.4%SourceNot directly comparable
skillsBench
Kimi K2.5 (Reasoning)—Qwen3.8 Max70.2%SourceNot directly comparable
Agents' Last Exam
Kimi K2.5 (Reasoning)—Qwen3.8 Max52.4%SourceNot directly comparable
AutomationBench
Kimi K2.5 (Reasoning)—Qwen3.8 Max27.3%SourceNot directly comparable
Toolathlon-Verified
Kimi K2.5 (Reasoning)—Qwen3.8 Max72.5%SourceNot directly comparable
WideResearch
Kimi K2.5 (Reasoning)—Qwen3.8 Max81.9%SourceNot directly comparable
HLE w/ tools
Kimi K2.5 (Reasoning)—Qwen3.8 Max56.2%SourceNot directly comparable
OSWorld-Verified
Kimi K2.5 (Reasoning)—Qwen3.8 Max86.1%SourceNot directly comparable
OSWorld 2.0
Kimi K2.5 (Reasoning)—Qwen3.8 Max19.4%SourceNot directly comparable
WebArena-Verified
Kimi K2.5 (Reasoning)—Qwen3.8 Max66.8%SourceNot directly comparable
AndroidWorld
Kimi K2.5 (Reasoning)—Qwen3.8 Max85.3%SourceNot directly comparable
MobileWorld
Kimi K2.5 (Reasoning)—Qwen3.8 Max77.8%SourceNot directly comparable
Coding
SWE-bench Verified
Kimi K2.5 (Reasoning)76.8%SourceQwen3.8 Max—Not directly comparable
Vibe Code Bench
Kimi K2.5 (Reasoning)17.54%SourceQwen3.8 Max—Not directly comparable
Terminal-Bench 2.1
Kimi K2.5 (Reasoning)—Qwen3.8 Max86.6%SourceNot directly comparable
SWE-bench Pro
Kimi K2.5 (Reasoning)—Qwen3.8 Max67.7%SourceNot directly comparable
deepSwe
Kimi K2.5 (Reasoning)—Qwen3.8 Max56.6%SourceNot directly comparable
NL2Repo
Kimi K2.5 (Reasoning)—Qwen3.8 Max55.9%SourceNot directly comparable
FrontierSWE
Kimi K2.5 (Reasoning)—Qwen3.8 Max73.5%SourceNot directly comparable
MLS-Bench Lite
Kimi K2.5 (Reasoning)—Qwen3.8 Max41.0%SourceNot directly comparable
PaperBench
Kimi K2.5 (Reasoning)—Qwen3.8 Max93.0%SourceNot directly comparable
Reasoning
Knowledge
GPQA
Kimi K2.5 (Reasoning)87.6%SourceQwen3.8 Max92.6%SourceQwen3.8 Max leads this result
MMLU-Pro
Kimi K2.5 (Reasoning)87.1%SourceQwen3.8 Max—Not directly comparable
GPQA-D
Kimi K2.5 (Reasoning)—Qwen3.8 Max92.6%SourceNot directly comparable
HLE
Kimi K2.5 (Reasoning)—Qwen3.8 Max43.6%SourceNot directly comparable
HLE w/o tools
Kimi K2.5 (Reasoning)—Qwen3.8 Max43.6%SourceNot directly comparable
Math
AIME 2025
Kimi K2.5 (Reasoning)96.1%SourceQwen3.8 Max—Not directly comparable
Multimodal
MMMU-Pro
Kimi K2.5 (Reasoning)78.5%SourceQwen3.8 Max82.3%SourceQwen3.8 Max leads this result
MathVision
Kimi K2.5 (Reasoning)—Qwen3.8 Max95.2%SourceNot directly comparable
MathVision w/ Python
Kimi K2.5 (Reasoning)—Qwen3.8 Max97.7%SourceNot directly comparable
BabyVision
Kimi K2.5 (Reasoning)—Qwen3.8 Max82.0%SourceNot directly comparable
BabyVision w/ Python
Kimi K2.5 (Reasoning)—Qwen3.8 Max91.3%SourceNot directly comparable
ZeroBench
Kimi K2.5 (Reasoning)—Qwen3.8 Max24.0%SourceNot directly comparable
ZeroBench w/ Python
Kimi K2.5 (Reasoning)—Qwen3.8 Max49.0%SourceNot directly comparable
MedXpertQA (MM)
Kimi K2.5 (Reasoning)—Qwen3.8 Max80.4%SourceNot directly comparable
ScreenSpot Pro
Kimi K2.5 (Reasoning)—Qwen3.8 Max84.5%SourceNot directly comparable
Vision2Web
Kimi K2.5 (Reasoning)—Qwen3.8 Max69.0%SourceNot directly comparable
CharXiv w/o tools
Kimi K2.5 (Reasoning)—Qwen3.8 Max88.4%SourceNot directly comparable
CharXiv
Kimi K2.5 (Reasoning)—Qwen3.8 Max93.5%SourceNot directly comparable
OmniDocBench 1.5
Kimi K2.5 (Reasoning)—Qwen3.8 Max92.1%SourceNot directly comparable
OCRBench V2
Kimi K2.5 (Reasoning)—Qwen3.8 Max74.2%SourceNot directly comparable
CC-OCR
Kimi K2.5 (Reasoning)—Qwen3.8 Max79.6%SourceNot directly comparable
RealWorldQA
Kimi K2.5 (Reasoning)—Qwen3.8 Max88.0%SourceNot directly comparable
ERQA
Kimi K2.5 (Reasoning)—Qwen3.8 Max77.8%SourceNot directly comparable
SimpleVQA
Kimi K2.5 (Reasoning)—Qwen3.8 Max75.0%SourceNot directly comparable
PerceptionBench
Kimi K2.5 (Reasoning)—Qwen3.8 Max63.5%SourceNot directly comparable
Video-MME (with subtitle)
Kimi K2.5 (Reasoning)—Qwen3.8 Max90.4%SourceNot directly comparable
VideoMMMU
Kimi K2.5 (Reasoning)—Qwen3.8 Max88.7%SourceNot directly comparable
MMVU
Kimi K2.5 (Reasoning)—Qwen3.8 Max82.4%SourceNot directly comparable
MLVU (M-Avg)
Kimi K2.5 (Reasoning)—Qwen3.8 Max90.8%SourceNot directly comparable
LVBench
Kimi K2.5 (Reasoning)—Qwen3.8 Max81.8%SourceNot directly comparable
Instruction following
IFBench
Kimi K2.5 (Reasoning)—Qwen3.8 Max82.8%SourceNot directly comparable
Frequently asked questions
Which is better, Kimi K2.5 (Reasoning) or Qwen3.8 Max?
Qwen3.8 Max has the higher public score estimate, 65.4 versus 58.61, but the 90% score intervals overlap. The higher estimate is not a decisive winner because the uncertainty ranges overlap.
Which is better for coding, Kimi K2.5 (Reasoning) or Qwen3.8 Max?
The published evidence does not provide a shared weighted coding basis for both models, so BenchLM does not name a coding winner.
Which is better for agentic tasks, Kimi K2.5 (Reasoning) or Qwen3.8 Max?
The published evidence does not provide a shared weighted agentic tasks basis for both models, so BenchLM does not name a agentic tasks winner.
Which costs less, Kimi K2.5 (Reasoning) 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, Kimi K2.5 (Reasoning) or Qwen3.8 Max?
Qwen3.8 Max has the larger documented context window: 1M, compared with 256K.
Related comparisons
Last updated August 3, 2026
Watch Kimi K2.5 (Reasoning) 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.