Coding work
Code generation, repair, and software-engineering tasks
Kimi K3
Kimi K3 has the higher public coding point estimate, 56.3 to 50.9, with Supported evidence for both models. The conditional ranges do not establish rank confidence.
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.
Kimi K3 has the higher public score estimate, 67 versus 63.37, but the 90% score intervals overlap. Treat that as a lead, not a settled winner. 33 results are shared. Category rows resting on Estimated evidence or different benchmark sets are marked directional and do not name a winner.
Kimi K3: $3 input / $15 output
Qwen3.8 Max: No comparable API price
Kimi K3: 1.05M
Qwen3.8 Max: 1M
Kimi K3: July 16, 2026
Qwen3.8 Max: August 3, 2026
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.
Code generation, repair, and software-engineering tasks
Kimi K3
Kimi K3 has the higher public coding point estimate, 56.3 to 50.9, with Supported evidence for both models. The conditional ranges do not establish rank confidence.
Tool use, computer use, and multi-step task completion
Kimi K3
Kimi K3 has the higher public agentic point estimate, 65.1 to 63.9, with Supported evidence for both models. The conditional ranges do not establish rank confidence.
Prompts that approach the documented context limit
Kimi K3
Kimi K3 has the larger documented context window.
1K fresh input + 500 output tokens
Not enough matched evidence
A complete comparable API-rate estimate is not available for both models.
200K cached + 20K fresh input + 10K output tokens
Not enough matched evidence
A complete comparable API-rate estimate is not available for both models.
50K fresh input + 3K output tokens
Not enough matched evidence
A complete comparable API-rate estimate is not available for both models.
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.
Like-for-like · 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.
The LLM Selector reads what you need a model for and returns a shortlist, ranked in the same order the job pages publish.
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
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.
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.
HLEKnowledge
Normalized gap 12.4AutomationBenchAgentic
Normalized gap 3.5CharXivMultimodal
Normalized gap 2.2GPQAKnowledge
Normalized gap 0.9LiveCodeBench (Vals)Coding
Normalized gap 0.7Each 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.
| Category | Kimi K3 | Qwen3.8 Max | Basis | Reading |
|---|---|---|---|---|
| Agentic | 65.1Supported · #21/140 | 63.9Supported · #24/140 | Like-for-likeBenchAlign v5.8 lane · 19 vs 19 public rows | Kimi K3 leads · intervals overlap |
| Coding | 56.3Supported · #23/163 | 50.9Supported · #45/163 | Like-for-likeBenchAlign v5.8 lane · 18 vs 15 public rows | Kimi K3 leads · intervals overlap |
| Knowledge | 67.5Supported · #20/181 | 65.2Supported · #29/181 | Like-for-likeBenchAlign v5.8 lane · 6 vs 6 public rows | Kimi K3 leads · intervals overlap |
| Reasoning | 65.4Supported · #21/161 | 61.1Estimated · #29/161 | Directional onlyBenchAlign v5.8 lane · 2 vs 2 public rows | Directional only |
| Multimodal | 66.7Estimated · #19/67 | 65.1Estimated · #24/67 | Directional onlyBenchAlign v5.8 lane · 14 vs 24 public rows | Directional only |
| Multilingual | Not ranked | Not ranked | Not comparableProvisional lane · 0 vs 0 weighted rows | Not comparable |
| Instruction following | Not ranked | 94.7#2/157 | Not comparableProvisional lane · 0 vs 1 weighted rows | Not comparable |
| Math | Not ranked | 72.6Unranked · 2 rankable rows | Not comparableProvisional lane · 0 vs 2 weighted rows | 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.
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.
1K fresh input + 500 output tokens
Qwen3.8 Max has no comparable published API token rate.
50K fresh input + 3K output tokens
Qwen3.8 Max has no comparable published API token rate.
200K cached + 20K fresh input + 10K output tokens
Qwen3.8 Max has no comparable published API token rate.
Overall score against output price per 1M tokens, one model per family. Pricing as of November 17, 2026.
Kimi K3
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 Kimi K3 on the full price vs performance chartSourced differences are shown directly. Missing facts stay explicit instead of being inferred from a model name or family.
Maximum documented context; output-token limits may be lower.
Kimi K3
1.05M
Qwen3.8 Max
Kimi K3
Not sourced
Qwen3.8 Max
qwen3.8-max
Alibaba Cloud Model Studio pricingA 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
Qwen3.8 Max
No comparable hosted API rate
Alibaba Cloud Model Studio pricingKimi K3
text, image, video
Kimi-K3 model cardQwen3.8 Max
text, image, video
Qwen Cloud model page for qwen3.8-maxKimi K3
Qwen3.8 Max
Kimi K3
Not sourced
Qwen3.8 Max
Not sourced
Kimi K3
Reasoning
Qwen3.8 Max
Reasoning
Kimi K3
Open Weight
Qwen3.8 Max
Open Weight
Kimi K3
Open Weight
Qwen3.8 Max
Open Weight
Kimi K3
2026-07-16
Qwen3.8 Max
2026-08-03
Kimi K3 has the higher public score estimate, 67 versus 63.37, but the 90% score intervals overlap. The higher estimate is not a decisive winner because the uncertainty ranges overlap.
Kimi K3 has the higher public coding point estimate, 56.3 to 50.9, with Supported evidence for both models. The conditional ranges do not establish rank confidence.
Kimi K3 has the higher public agentic tasks point estimate, 65.1 to 63.9, with Supported evidence for both models. The conditional ranges do not establish rank confidence.
Both models do not have comparable published API token rates, so this page does not name a universal price winner.
Kimi K3 has the larger documented context window: 1.05M, compared with 1M.
Estimates at 50,000 req/day · 1000 tokens/req average.
The full public result ledger is available for audit without forcing a wide desktop table onto a phone.
Terminal-Bench 4.1
Not directly comparable
Finch
Not directly comparable
b3 Agent Security
Not directly comparable
Cyber Misuse Refusal
Not directly comparable
Terminal-Bench 2.1
Kimi K3 leads this result
BrowseComp
Not directly comparable
DeepSearchQA
Not directly comparable
Toolathlon-Verified
Kimi K3 leads this result
MCP Atlas
Not directly comparable
AutomationBench
Kimi K3 leads this result
JobBench
Qwen3.8 Max leads this result
APEX-Agents
Not directly comparable
SpreadsheetBench 2
Not directly comparable
DECK-Bench
Not directly comparable
Terminal-Bench 2.1 (Vals)
Kimi K3 leads this result
ApprenticeBench
Not directly comparable
Hermes Index
Shared sourceKimi K3 leads this result
Hermes Bench
Shared sourceKimi K3 leads this result
MazeBench
Kimi K3 leads this result
CoWorkBench
Not directly comparable
skillsBench
Not directly comparable
Agents' Last Exam
Not directly comparable
WideResearch
Not directly comparable
HLE w/ tools
Not directly comparable
OSWorld-Verified
Not directly comparable
OSWorld 2.0
Not directly comparable
WebArena-Verified
Not directly comparable
AndroidWorld
Not directly comparable
MobileWorld
Not directly comparable
CWE-bench
Not directly comparable
ScarfBench
Shared sourceQwen3.8 Max leads this result
DeepSWE
Kimi K3 leads this result
CursorBench 3.2
Not directly comparable
FrontierSWE
Kimi K3 leads this result
ProgramBench
Not directly comparable
Kimi Code Bench v2
Not directly comparable
sweMarathon
Not directly comparable
PostTrainBench (lab runs)
Not directly comparable
MLS-Bench Lite
Kimi K3 leads this result
VulcanBench v3
Qwen3.8 Max leads this result
OpenHarmony Bench v1.0
Shared sourceQwen3.8 Max leads this result
FrontierSWE v2
Shared sourceKimi K3 leads this result
LiveCodeBench (Vals)
Qwen3.8 Max leads this result
SWE-bench (Vals)
Kimi K3 leads this result
PostTrainBench v1.1
Not directly comparable
Bug Hunt Bench
Shared sourceQwen3.8 Max leads this result
FrontierCode 2.0
Not directly comparable
PostTrainBench
Not directly comparable
Terminal-Bench 2.1
Not directly comparable
SWE-bench Pro
Not directly comparable
NL2Repo
Not directly comparable
PaperBench
Not directly comparable
OpenHarmony Bench v1.2
Not directly comparable
ExtractBench (LlamaIndex)
Not directly comparable
OfficeQA Pro
Not directly comparable
MMMU-Pro
Qwen3.8 Max leads this result
MMMU-Pro w/ Python
Not directly comparable
CharXiv w/o tools
Qwen3.8 Max leads this result
CharXiv
Qwen3.8 Max leads this result
MathVision
Qwen3.8 Max leads this result
MathVision w/ Python
Kimi K3 leads this result
BabyVision w/ Python
Qwen3.8 Max leads this result
ZeroBench
Qwen3.8 Max leads this result
ZeroBench w/ Python
Qwen3.8 Max leads this result
WorldVQA ForceAnswer
Not directly comparable
OmniDocBench
Not directly comparable
PerceptionBench
Qwen3.8 Max leads this result
BabyVision
Not directly comparable
MedXpertQA (MM)
Not directly comparable
ScreenSpot Pro
Not directly comparable
Vision2Web
Not directly comparable
OmniDocBench 1.5
Not directly comparable
OCRBench V2
Not directly comparable
CC-OCR
Not directly comparable
RealWorldQA
Not directly comparable
ERQA
Not directly comparable
SimpleVQA
Not directly comparable
Video-MME (with subtitle)
Not directly comparable
VideoMMMU
Not directly comparable
MMVU
Not directly comparable
MLVU (M-Avg)
Not directly comparable
LVBench
Not directly comparable
GPQA
Kimi K3 leads this result
GPQA-D
Kimi K3 leads this result
HLE
Kimi K3 leads this result
HLE w/o tools
Qwen3.8 Max leads this result
GPQA Diamond (Vals)
Qwen3.8 Max leads this result
MMLU-Pro (Vals)
Qwen3.8 Max leads this result
One weekly email when material rank, price, or benchmark evidence changes make this matchup worth revisiting.
Read a sample issueJoin 7,000+ subscribers.
Last updated November 17, 2026