Chat turn cost
1K fresh input + 500 output tokens
Kimi K3.1 Pro
Kimi K3.1 Pro has the lower estimated token cost for this stated workload. Costs use the listed standard API rates.
Confidence: listed-rates
Kimi K3.1 Pro vs Kimi K3
Updated October 23, 2026. We do not rank this pair: at least one has no public score. Public scores include evidence status and uncertainty.
Share or export
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. 18 results are shared. Category rows resting on Estimated evidence or different benchmark sets are marked directional and do not name a winner.
- Shared results
- 18
- Kimi K3.1 Pro only
- 24
- Kimi K3 only
- 35
- Like-for-like categories
- 1 / 8
Supported: Kimi K3How 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.
Cache-heavy agent loop cost
200K cached + 20K fresh input + 10K output tokens
Kimi K3.1 Pro
Kimi K3.1 Pro 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
Kimi K3.1 Pro
Kimi K3.1 Pro has the lower estimated token cost for this stated workload. Costs use the listed standard API rates.
Confidence: listed-ratesCoding work
Code generation, repair, and software-engineering tasks
Not enough matched evidence
Kimi K3.1 Pro is not ranked on the public lane for coding, so no winner is named for coding.
Confidence: limitedAgentic work
Tool use, computer use, and multi-step task completion
Not enough matched evidence
Kimi K3.1 Pro is not ranked on the public lane for agentic, so no winner is named for agentic.
Confidence: limitedLong documents
Prompts that approach the documented context limit
No clear pick
The documented context windows are equal.
Confidence: documented
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.
—Kimi K3.1 Pro58.8Kimi K3
Not comparable · BenchAlign v5.8
The coding row is not comparable on the public lane: at least one model is not measured or not ranked there.
Coding scores combine specific tasks and setups. Match the editor, harness, and effort to your workflow.
Same basis rules as the category table below
What is actually comparable
Shared results can support a head-to-head reading. Results present for only one model describe coverage, not superiority.
24 Kimi K3.1 Pro only
18 Shared
35 Kimi K3 only
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.
HLEKnowledge
Normalized gap 14.0Kimi K3.1 Pro:70%Kimi K3:56%HLE w/o toolsKnowledge
Normalized gap 7.6Kimi K3.1 Pro:51.1%Kimi K3:43.5%OfficeQA ProMultimodal
Normalized gap 4.1Kimi K3.1 Pro:67.4%Kimi K3:63.3%MMMU-ProMultimodal
Normalized gap 2.5Kimi K3.1 Pro:84.1%Kimi K3:81.6%CharXivMultimodal
Normalized gap 2.1Kimi K3.1 Pro:93.4%Kimi K3:91.3%
Bars run 0–100 on each benchmark’s normalized display scale
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, 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- Kimi K3.1 Pro
- 98.0
- #1/56
- Kimi K3
- 90.2
- #5/56
- Basis
- Provisional lane · 3 vs 3 weighted rows
- Reading
- Kimi K3.1 Pro leads
Agentic
Not comparable- Kimi K3.1 Pro
- Not ranked
- Kimi K3
- 69.3
- Supported · #10/132
- Basis
- BenchAlign v5.8 lane · 16 vs 15 public rows
- Reading
- Not comparable
Coding
Not comparable- Kimi K3.1 Pro
- Not ranked
- Kimi K3
- 58.8
- Supported · #23/153
- Basis
- BenchAlign v5.8 lane · 6 vs 15 public rows
- Reading
- Not comparable
Reasoning
Not comparable- Kimi K3.1 Pro
- Not ranked
- Kimi K3
- 69.8
- #18/27
- Basis
- Provisional lane · 0 vs 1 weighted rows
- Reading
- Not comparable
Knowledge
Not comparable- Kimi K3.1 Pro
- Not ranked
- Kimi K3
- 67.5
- Supported · #24/187
- Basis
- BenchAlign v5.8 lane · 5 vs 6 public rows
- Reading
- Not comparable
Multilingual
Not comparable- Kimi K3.1 Pro
- Not ranked
- Kimi K3
- Not ranked
- Basis
- Provisional lane · 0 vs 0 weighted rows
- Reading
- Not comparable
Instruction following
Not comparable- Kimi K3.1 Pro
- Not ranked
- Kimi K3
- Not ranked
- Basis
- Provisional lane · 0 vs 0 weighted rows
- Reading
- Not comparable
Math
Not comparable- Kimi K3.1 Pro
- Not ranked
- Kimi K3
- Not ranked
- Basis
- Provisional lane · 0 vs 0 weighted rows
- Reading
- Not comparable
| Category | Kimi K3.1 Pro | Kimi K3 | Basis | Reading |
|---|---|---|---|---|
| Multimodal | 98.0#1/56 | 90.2#5/56 | Like-for-likeProvisional lane · 3 vs 3 weighted rows | Kimi K3.1 Pro leads |
| Agentic | Not ranked | 69.3Supported · #10/132 | Not comparableBenchAlign v5.8 lane · 16 vs 15 public rows | Not comparable |
| Coding | Not ranked | 58.8Supported · #23/153 | Not comparableBenchAlign v5.8 lane · 6 vs 15 public rows | Not comparable |
| Reasoning | Not ranked | 69.8#18/27 | Not comparableProvisional lane · 0 vs 1 weighted rows | Not comparable |
| Knowledge | Not ranked | 67.5Supported · #24/187 | Not comparableBenchAlign v5.8 lane · 5 vs 6 public rows | Not comparable |
| Multilingual | Not ranked | Not ranked | Not comparableProvisional lane · 0 vs 0 weighted rows | Not comparable |
| Instruction following | Not ranked | Not ranked | Not comparableProvisional lane · 0 vs 0 weighted rows | Not comparable |
| Math | Not ranked | Not ranked | Not comparableProvisional lane · 0 vs 0 weighted rows | Not comparable |
Ranks count the models scored in each category’s lane, so the agentic, coding, 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 K3.1 Pro
- $0.007
- Fits in one request
- Kimi K3
- $0.0105
- Fits in one request
Kimi K3.1 Pro has the lower modeled cost
Costs use the listed standard API rates.
Repository review
50K fresh input + 3K output tokens
- Kimi K3.1 Pro
- $0.13
- Fits in one request
- Kimi K3
- $0.195
- Fits in one request
Kimi K3.1 Pro 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 K3.1 Pro
- $0.18
- Fits in one request
- Kimi K3
- $0.27
- Fits in one request
Kimi K3.1 Pro has the lower modeled cost
Costs use the listed standard API rates.
Cached input falls back to the list input rate only where a cached rate is unpublished
Specification differences
Sourced differences are shown directly. Missing facts stay explicit instead of being inferred from a model name or family.
SpecificationKimi K3.1 ProKimi K3
Context window
Maximum documented context; output-token limits may be lower.
Kimi K3.1 Pro
1.05M
Kimi K3
1.05M
API model ID
Kimi K3.1 Pro
Not sourced
Kimi K3
Not sourced
Cached-input rate
A missing cached-input rate falls back to the listed input rate only in the stated workload estimate.
Kimi K3.1 Pro
$0.2 per 1M cached input tokens
Kimi K3
$0.3 per 1M cached input tokens
Documented inputs
Kimi K3.1 Pro
Not sourced
Kimi K3
Not sourced
Documented outputs
Kimi K3.1 Pro
Not sourced
Kimi K3
Not sourced
Provider availability
Kimi K3.1 Pro
Not sourced
Kimi K3
Not sourced
Reasoning profile
Kimi K3.1 Pro
Reasoning
Kimi K3
Reasoning
Weight access
Kimi K3.1 Pro
Pending
Kimi K3
Pending
License
Kimi K3.1 Pro
Pending
Kimi K3
Pending
Release date
Kimi K3.1 Pro
2026-10-23
Kimi K3
2026-07-16
If you are considering the documented upgrade path
- Deployment change
- Both entries list Moonshot AI as the provider. Confirm endpoint, model ID, limits, and feature support before switching.
- 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
- Repository review: $0.13 vs $0.195. Cache-heavy agent loop: $0.18 vs $0.27.
- Context tradeoff
- Both models list 1.05M.
Run the same representative tasks against both endpoints before changing production traffic.
Questions
Which is better, Kimi K3.1 Pro or Kimi K3?
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, Kimi K3.1 Pro or Kimi K3?
Kimi K3.1 Pro is not ranked on the public lane for coding, so no winner is named for coding.
Which is better for agentic tasks, Kimi K3.1 Pro or Kimi K3?
Kimi K3.1 Pro is not ranked on the public lane for agentic tasks, so no winner is named for agentic tasks.
Which costs less, Kimi K3.1 Pro or Kimi K3?
For the stated presets, chat costs $0.007 on Kimi K3.1 Pro and $0.0105 on Kimi K3; repository review costs $0.13 and $0.195; the cache-heavy agent loop costs $0.18 and $0.27. Costs use the listed standard API rates.
Which has the larger context window, Kimi K3.1 Pro or Kimi K3?
Both models list the same context window, 1.05M.
Benchmark evidence
The full public result ledger is available for audit without forcing a wide desktop table onto a phone.
Browse raw public benchmark evidence77 rows
Agentic
Terminal-Bench 4.0
Kimi K3.1 Pro65.70%SourceKimi K3—Not directly comparable
Terminal-Bench-Science 0.1
Kimi K3.1 Pro38.1%SourceKimi K3—Not directly comparable
HLE w/ tools
Kimi K3.1 Pro70.0%SourceKimi K3—Not directly comparable
BrowseComp
Kimi K3.1 Pro92.1%SourceKimi K391.2%SourceKimi K3.1 Pro leads this result
BioMysteryBench
Kimi K3.1 Pro80.6%SourceKimi K3—Not directly comparable
Agents' Last Exam
Kimi K3.1 Pro43.2%SourceKimi K3—Not directly comparable
Agents' Last Exam (CLI)
Kimi K3.1 Pro44.2%SourceKimi K3—Not directly comparable
APEX-Agents
Kimi K3.1 Pro67.2%SourceKimi K337.6%SourceKimi K3.1 Pro leads this result
Frontier Finance
Kimi K3.1 Pro58.6%SourceKimi K3—Not directly comparable
HANDBOOK.md
Kimi K3.1 Pro44.6%SourceKimi K3—Not directly comparable
OSWorld 2.0
Kimi K3.1 Pro79.6%SourceKimi K3—Not directly comparable
Business Operation
Kimi K3.1 Pro37.8%SourceKimi K3—Not directly comparable
Financial Modeling
Kimi K3.1 Pro57.6%SourceKimi K3—Not directly comparable
Financial Research
Kimi K3.1 Pro70.4%SourceKimi K3—Not directly comparable
Kimi Hallu Bench Hard (tools)
Kimi K3.1 Pro8.3%SourceKimi K3—Not directly comparable
ExploitGym
Kimi K3.1 Pro20.7%SourceKimi K3—Not directly comparable
Terminal-Bench 2.1
Kimi K3.1 Pro—Kimi K388.3%SourceNot directly comparable
DeepSearchQA
Kimi K3.1 Pro—Kimi K395.0%SourceNot directly comparable
Toolathlon-Verified
Kimi K3.1 Pro—Kimi K373.2%SourceNot directly comparable
MCP Atlas
Kimi K3.1 Pro—Kimi K384.2%SourceNot directly comparable
AutomationBench
Kimi K3.1 Pro—Kimi K330.8%SourceNot directly comparable
JobBench
Kimi K3.1 Pro—Kimi K352.9%SourceNot directly comparable
SpreadsheetBench 2
Kimi K3.1 Pro—Kimi K334.8%SourceNot directly comparable
DECK-Bench
Kimi K3.1 Pro—Kimi K373.5%SourceNot directly comparable
Terminal-Bench 2.1 (Vals)
Kimi K3.1 Pro—Kimi K380.9%SourceNot directly comparable
ApprenticeBench
Kimi K3.1 Pro—Kimi K318%SourceNot directly comparable
Hermes Index
Kimi K3.1 Pro—Kimi K339.2%SourceNot directly comparable
Hermes Bench
Kimi K3.1 Pro—Kimi K371.5%SourceNot directly comparable
MazeBench
Kimi K3.1 Pro—Kimi K31%SourceNot directly comparable
Coding
DeepSWE
Kimi K3.1 Pro75.5%SourceKimi K367.5%SourceKimi K3.1 Pro leads this result
ProgramBench (binary)
Kimi K3.1 Pro31.5%SourceKimi K3—Not directly comparable
Kimi Code Bench v2
Kimi K3.1 Pro77.1%SourceKimi K372.9%SourceKimi K3.1 Pro leads this result
Kimi Code Bench 3.0 Long-horizon
Kimi K3.1 Pro61.2%SourceKimi K3—Not directly comparable
sweMarathon
Kimi K3.1 Pro51.7%SourceKimi K342%SourceKimi K3.1 Pro leads this result
PostTrainBench v1.2
Kimi K3.1 Pro36.5%SourceKimi K3—Not directly comparable
CursorBench 3.2
Kimi K3.1 Pro—Kimi K360.8%SourceNot directly comparable
FrontierSWE
Kimi K3.1 Pro—Kimi K381.2%SourceNot directly comparable
ProgramBench
Kimi K3.1 Pro—Kimi K377.8%SourceNot directly comparable
PostTrain Bench
Kimi K3.1 Pro—Kimi K336.6%SourceNot directly comparable
MLS-Bench Lite
Kimi K3.1 Pro—Kimi K348.3%SourceNot directly comparable
VulcanBench v3
Kimi K3.1 Pro—Kimi K373.7%SourceNot directly comparable
OpenHarmony Bench
Kimi K3.1 Pro—Kimi K357.3%SourceNot directly comparable
FrontierSWE v2
Kimi K3.1 Pro—Kimi K325.9%SourceNot directly comparable
LiveCodeBench (Vals)
Kimi K3.1 Pro—Kimi K387.2%SourceNot directly comparable
SWE-bench (Vals)
Kimi K3.1 Pro—Kimi K393.4%SourceNot directly comparable
PostTrainBench v1.1
Kimi K3.1 Pro—Kimi K332.0%SourceNot directly comparable
Bug Hunt Bench
Kimi K3.1 Pro—Kimi K321.0 fixesSourceNot directly comparable
Reasoning
Multimodal
OfficeQA Pro
Kimi K3.1 Pro67.4%SourceKimi K363.3%SourceKimi K3.1 Pro leads this result
OfficeQA Pro v2
Kimi K3.1 Pro64.4%SourceKimi K3—Not directly comparable
BenchCAD Vision2Code (no tools)
Kimi K3.1 Pro0.728SourceKimi K3—Not directly comparable
BenchCAD Vision2Code (tools)
Kimi K3.1 Pro0.927SourceKimi K3—Not directly comparable
MMMU-Pro
Kimi K3.1 Pro84.1%SourceKimi K381.6%SourceKimi K3.1 Pro leads this result
MMMU-Pro w/ Python
Kimi K3.1 Pro90%SourceKimi K383.4%SourceKimi K3.1 Pro leads this result
CharXiv
Kimi K3.1 Pro93.4%SourceKimi K391.3%SourceKimi K3.1 Pro leads this result
CharXiv w/o tools
Kimi K3.1 Pro89%SourceKimi K384.8%SourceKimi K3.1 Pro leads this result
Chartography (tools)
Kimi K3.1 Pro90.2%SourceKimi K3—Not directly comparable
BabyVision w/ Python
Kimi K3.1 Pro95.4%SourceKimi K385.7%SourceKimi K3.1 Pro leads this result
ZeroBench
Kimi K3.1 Pro39.0%SourceKimi K323.0%SourceKimi K3.1 Pro leads this result
ZeroBench w/ Python
Kimi K3.1 Pro63.0%SourceKimi K341.0%SourceKimi K3.1 Pro leads this result
GDP.pdf (Gemini judge)
Kimi K3.1 Pro34.6%SourceKimi K3—Not directly comparable
GDP.pdf w/ Python (Gemini judge)
Kimi K3.1 Pro85.7%SourceKimi K3—Not directly comparable
PerceptionBench
Kimi K3.1 Pro68.3%SourceKimi K358.5%SourceKimi K3.1 Pro leads this result
ExtractBench (LlamaIndex)
Kimi K3.1 Pro—Kimi K383.17%SourceNot directly comparable
MathVision
Kimi K3.1 Pro—Kimi K394.3%SourceNot directly comparable
MathVision w/ Python
Kimi K3.1 Pro—Kimi K397.8%SourceNot directly comparable
WorldVQA ForceAnswer
Kimi K3.1 Pro—Kimi K351.0%SourceNot directly comparable
OmniDocBench
Kimi K3.1 Pro—Kimi K391.1%SourceNot directly comparable
Knowledge
GPQA
Kimi K3.1 Pro94.2%SourceKimi K393.5%SourceKimi K3.1 Pro leads this result
GPQA-D
Kimi K3.1 Pro94.2%SourceKimi K393.5%SourceKimi K3.1 Pro leads this result
HLE
Kimi K3.1 Pro70%SourceKimi K356%SourceKimi K3.1 Pro leads this result
HLE w/o tools
Kimi K3.1 Pro51.1%SourceKimi K343.5%SourceKimi K3.1 Pro leads this result
Kimi Hallu Bench
Kimi K3.1 Pro5.4%SourceKimi K3—Not directly comparable
GPQA Diamond (Vals)
Kimi K3.1 Pro—Kimi K392.9%SourceNot directly comparable
MMLU-Pro (Vals)
Kimi K3.1 Pro—Kimi K388.0%SourceNot directly comparable
Instruction following
Gray Swan IPI (15 attempts)
Kimi K3.1 Pro—Kimi K352.7%SourceNot directly comparable
77 public results · 18 shared
Watch Kimi K3.1 Pro vs Kimi K3
One weekly email when material rank, price, or benchmark evidence changes make this matchup worth revisiting.
Read a sample issueJoin 5,600+ readers.
Last updated October 23, 2026