Agentic
Like-for-like- DeepSeek V4 Pro 0813
- 56.1
- Supported · #37/151
- Kimi K3
- 71.9
- Supported · #4/151
- Basis
- BenchAlign lane · 11 vs 11 public rows
- Reading
- Kimi K3 leads
DeepSeek V4 Pro 0813 vs Kimi K3
Updated September 4, 2026. Public scores include evidence status and uncertainty. They are not guarantees for a specific workload.
Decision reading
Kimi K3 has the higher public score estimate, 74.87 versus 66.35, but the 90% score intervals overlap. Treat that as a lead, not a settled winner.
14 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
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
Kimi K3
Kimi K3 leads on the public coding lane, 68 to 52.2, with Supported evidence for both models, although the 90% intervals overlap.
Confidence: limited
Agentic work
Tool use, computer use, and multi-step task completion
Kimi K3
Kimi K3 leads on the public agentic lane, 71.9 to 56.1, with Supported evidence for both models and non-overlapping 90% intervals.
Confidence: stronger
Long documents
Prompts that approach the documented context limit
Kimi K3
Kimi K3 has the larger documented context window.
Confidence: documented
Show secondary and unsupported calls
Chat turn cost
1K fresh input + 500 output tokens
DeepSeek V4 Pro 0813
DeepSeek V4 Pro 0813 has the lower estimated token cost for this stated workload. Costs use the listed standard API rates.
Confidence: listed-rates
Cache-heavy agent loop cost
200K cached + 20K fresh input + 10K output tokens
DeepSeek V4 Pro 0813
DeepSeek V4 Pro 0813 has the lower estimated token cost for this stated workload. Costs use the listed standard API rates.
Confidence: listed-rates
Repository review cost
50K fresh input + 3K output tokens
DeepSeek V4 Pro 0813
DeepSeek V4 Pro 0813 has the lower estimated token cost for this stated workload. Costs use the listed standard API rates.
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.
25 DeepSeek V4 Pro 0813 only
14 Shared
29 Kimi K3 only
- Shared results
- 14
- DeepSeek V4 Pro 0813 only
- 25
- Kimi K3 only
- 29
- 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.
Coding
Like-for-like- DeepSeek V4 Pro 0813
- 52.2
- Supported · #58/183
- Kimi K3
- 68.0
- Supported · #7/183
- Basis
- BenchAlign lane · 14 vs 13 public rows
- Reading
- Kimi K3 leads · intervals overlap
Reasoning
Directional only- DeepSeek V4 Pro 0813
- 58.7
- #18/22
- Kimi K3
- 78.5
- #3/22
- Basis
- Provisional lane · 0 vs 0 weighted rows
- Reading
- Directional only
Knowledge
Directional only- DeepSeek V4 Pro 0813
- 60.9
- Estimated · #37/181
- Kimi K3
- 72.3
- Supported · #8/181
- Basis
- BenchAlign lane · 8 vs 6 public rows
- Reading
- Directional only
Math
Not comparable- DeepSeek V4 Pro 0813
- 80.4
- Unranked · 4 rankable rows
- Kimi K3
- Not ranked
- Basis
- Provisional lane · 1 vs 0 weighted rows
- Reading
- Not comparable
Multilingual
Not comparable- DeepSeek V4 Pro 0813
- Not ranked
- Kimi K3
- Not ranked
- Basis
- Provisional lane · 0 vs 0 weighted rows
- Reading
- Not comparable
Multimodal
Not comparable- DeepSeek V4 Pro 0813
- Not ranked
- Kimi K3
- 89.5
- #1/48
- Basis
- Provisional lane · 0 vs 3 weighted rows
- Reading
- Not comparable
Instruction following
Not comparable- DeepSeek V4 Pro 0813
- Not ranked
- Kimi K3
- Not ranked
- Basis
- Provisional lane · 0 vs 0 weighted rows
- Reading
- Not comparable
| Category | DeepSeek V4 Pro 0813 | Kimi K3 | Basis | Reading |
|---|---|---|---|---|
| Agentic | 56.1Supported · #37/151 | 71.9Supported · #4/151 | Like-for-likeBenchAlign lane · 11 vs 11 public rows | Kimi K3 leads |
| Coding | 52.2Supported · #58/183 | 68.0Supported · #7/183 | Like-for-likeBenchAlign lane · 14 vs 13 public rows | Kimi K3 leads · intervals overlap |
| Reasoning | 58.7#18/22 | 78.5#3/22 | Directional onlyProvisional lane · 0 vs 0 weighted rows | Directional only |
| Knowledge | 60.9Estimated · #37/181 | 72.3Supported · #8/181 | Directional onlyBenchAlign lane · 8 vs 6 public rows | Directional only |
| Math | 80.4Unranked · 4 rankable rows | Not ranked | Not comparableProvisional lane · 1 vs 0 weighted rows | Not comparable |
| Multilingual | Not ranked | Not ranked | Not comparableProvisional lane · 0 vs 0 weighted rows | Not comparable |
| Multimodal | Not ranked | 89.5#1/48 | Not comparableProvisional lane · 0 vs 3 weighted rows | Not comparable |
| Instruction following | 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) 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.
Terminal-Bench 2.0
Agentic
DeepSeek V4 Pro 0813: 67.9%Kimi K3: 88.3%Normalized gap 20.4DeepSeek V4 Pro 0813 source Kimi K3 sourceHLE
Knowledge
DeepSeek V4 Pro 0813: 42.7%Kimi K3: 56%Normalized gap 13.3DeepSeek V4 Pro 0813 source Kimi K3 sourceBrowseComp
Agentic
DeepSeek V4 Pro 0813: 83.4%Kimi K3: 91.2%Normalized gap 7.8DeepSeek V4 Pro 0813 source Kimi K3 sourceGPQA
Knowledge
DeepSeek V4 Pro 0813: 90.1%Kimi K3: 93.5%Normalized gap 3.4DeepSeek V4 Pro 0813 source Kimi K3 sourceAutomationBench
Agentic
DeepSeek V4 Pro 0813: 31.8%Kimi K3: 30.8%Normalized gap 1.0DeepSeek V4 Pro 0813 source Kimi K3 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
- DeepSeek V4 Pro 0813
- $0.00087
- Fits in one request
- Kimi K3
- $0.0105
- Fits in one request
DeepSeek V4 Pro 0813 has the lower modeled cost
Costs use the listed standard API rates.
Repository review
50K fresh input + 3K output tokens
- DeepSeek V4 Pro 0813
- $0.02436
- Fits in one request
- Kimi K3
- $0.195
- Fits in one request
DeepSeek V4 Pro 0813 has the lower modeled cost
Costs use the listed standard API rates.
Cache-heavy agent loop
200K cached + 20K fresh input + 10K output tokens
- DeepSeek V4 Pro 0813
- $0.01812
- Fits in one request
- Kimi K3
- $0.27
- Fits in one request
DeepSeek V4 Pro 0813 has the lower modeled cost
Costs use the listed standard API rates.
Specification differences
Sourced differences are shown directly. Missing facts stay explicit instead of being inferred from a model name or family.
SpecificationDeepSeek V4 Pro 0813Kimi K3
Context window
Maximum documented context; output-token limits may be lower.
DeepSeek V4 Pro 0813
Kimi K3
1.05M
API model ID
DeepSeek V4 Pro 0813
deepseek-v4-pro
DeepSeek models and pricingKimi K3
Not sourced
Cached-input rate
A missing cached-input rate falls back to the listed input rate only in the stated workload estimate.
DeepSeek V4 Pro 0813
$0.003625 per 1M cached input tokens
Kimi K3
$0.3 per 1M cached input tokens
Documented inputs
DeepSeek V4 Pro 0813
Kimi K3
Not sourced
Documented outputs
DeepSeek V4 Pro 0813
Kimi K3
Not sourced
Provider availability
DeepSeek V4 Pro 0813
Generally Available · DeepSeek API
DeepSeek models and pricingKimi K3
Not sourced
Reasoning profile
DeepSeek V4 Pro 0813
Reasoning
Kimi K3
Reasoning
Weight access
DeepSeek V4 Pro 0813
Proprietary
Kimi K3
Pending
License
DeepSeek V4 Pro 0813
Proprietary
Kimi K3
Pending
Release date
DeepSeek V4 Pro 0813
2026-08-13
Kimi K3
2026-07-16
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
- Kimi K3 has the higher public score estimate, 74.87 versus 66.35, but the 90% score intervals overlap.
- Workload cost
- Repository review: $0.02436 vs $0.195. Cache-heavy agent loop: $0.01812 vs $0.27.
- Context tradeoff
- Kimi K3 has the larger documented window (1.05M).
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 evidence68 rows
Agentic
Terminal-Bench 2.0
DeepSeek V4 Pro 081367.9%SourceKimi K388.3%SourceKimi K3 leads this result
Terminal-Bench 2.1
DeepSeek V4 Pro 081387.9%SourceKimi K3—Not directly comparable
BrowseComp
DeepSeek V4 Pro 081383.4%SourceKimi K391.2%SourceKimi K3 leads this result
HLE w/ tools
DeepSeek V4 Pro 081360.0%SourceKimi K3—Not directly comparable
MCP Atlas
DeepSeek V4 Pro 081373.6%SourceKimi K384.2%SourceKimi K3 leads this result
Toolathlon
DeepSeek V4 Pro 081351.8%SourceKimi K3—Not directly comparable
CyberGym
DeepSeek V4 Pro 081383.3%SourceKimi K3—Not directly comparable
Toolathlon-Verified
DeepSeek V4 Pro 081374.1%SourceKimi K373.2%SourceDeepSeek V4 Pro 0813 leads this result
Agents' Last Exam
DeepSeek V4 Pro 081325.7%SourceKimi K3—Not directly comparable
AutomationBench
DeepSeek V4 Pro 081331.8%SourceKimi K330.8%SourceDeepSeek V4 Pro 0813 leads this result
Terminal-Bench 2.1 (Vals)
DeepSeek V4 Pro 081354.7%SourceKimi K380.9%SourceKimi K3 leads this result
DeepSearchQA
DeepSeek V4 Pro 0813—Kimi K395.0%SourceNot directly comparable
JobBench
DeepSeek V4 Pro 0813—Kimi K352.9%SourceNot directly comparable
APEX-Agents
DeepSeek V4 Pro 0813—Kimi K337.6%SourceNot directly comparable
SpreadsheetBench 2
DeepSeek V4 Pro 0813—Kimi K334.8%SourceNot directly comparable
DECK-Bench
DeepSeek V4 Pro 0813—Kimi K373.5%SourceNot directly comparable
Coding
LiveCodeBench Pass@1-COT
DeepSeek V4 Pro 081393.5%SourceKimi K3—Not directly comparable
Codeforces
DeepSeek V4 Pro 08133206.0SourceKimi K3—Not directly comparable
SWE-bench Verified
DeepSeek V4 Pro 081380.6%SourceKimi K3—Not directly comparable
SWE-bench Pro
DeepSeek V4 Pro 081355.4%SourceKimi K3—Not directly comparable
SWE Multilingual
DeepSeek V4 Pro 081376.2%SourceKimi K3—Not directly comparable
Terminal-Bench 2.0
DeepSeek V4 Pro 081367.9%SourceKimi K3—Not directly comparable
Vibe Code Bench
DeepSeek V4 Pro 081349.93%SourceKimi K3—Not directly comparable
Terminal-Bench 2.1
DeepSeek V4 Pro 081387.9%SourceKimi K3—Not directly comparable
NL2Repo
DeepSeek V4 Pro 081361.5%SourceKimi K3—Not directly comparable
deepSwe
DeepSeek V4 Pro 081362.7%SourceKimi K367.5%SourceKimi K3 leads this result
DSBench-FullStack
DeepSeek V4 Pro 081371.1%SourceKimi K3—Not directly comparable
DSBench-Hard
DeepSeek V4 Pro 081367.2%SourceKimi K3—Not directly comparable
LiveCodeBench (Vals)
DeepSeek V4 Pro 081387.5%SourceKimi K387.2%SourceDeepSeek V4 Pro 0813 leads this result
SWE-bench (Vals)
DeepSeek V4 Pro 081396.4%SourceKimi K393.4%SourceDeepSeek V4 Pro 0813 leads this result
cursorBench32
DeepSeek V4 Pro 0813—Kimi K360.8%SourceNot directly comparable
FrontierSWE
DeepSeek V4 Pro 0813—Kimi K381.2%SourceNot directly comparable
ProgramBench
DeepSeek V4 Pro 0813—Kimi K377.8%SourceNot directly comparable
Kimi Code Bench v2
DeepSeek V4 Pro 0813—Kimi K372.9%SourceNot directly comparable
sweMarathon
DeepSeek V4 Pro 0813—Kimi K342%SourceNot directly comparable
PostTrain Bench
DeepSeek V4 Pro 0813—Kimi K336.6%SourceNot directly comparable
MLS-Bench Lite
DeepSeek V4 Pro 0813—Kimi K348.3%SourceNot directly comparable
VulcanBench v3
DeepSeek V4 Pro 0813—Kimi K373.7%SourceNot directly comparable
OpenHarmony Bench
DeepSeek V4 Pro 0813—Kimi K357.3%SourceNot directly comparable
FrontierSWE v2
DeepSeek V4 Pro 0813—Kimi K325.9%SourceNot directly comparable
Reasoning
Knowledge
MMLU-Pro
DeepSeek V4 Pro 081387.5%SourceKimi K3—Not directly comparable
SimpleQA
DeepSeek V4 Pro 081357.9%SourceKimi K3—Not directly comparable
Chinese-SimpleQA
DeepSeek V4 Pro 081384.4%SourceKimi K3—Not directly comparable
GPQA
DeepSeek V4 Pro 081390.1%SourceKimi K393.5%SourceKimi K3 leads this result
GPQA-D
DeepSeek V4 Pro 081390.1%SourceKimi K393.5%SourceKimi K3 leads this result
HLE
DeepSeek V4 Pro 081342.7%SourceKimi K356%SourceKimi K3 leads this result
GPQA Diamond (Vals)
DeepSeek V4 Pro 081392.4%SourceKimi K392.9%SourceKimi K3 leads this result
MMLU-Pro (Vals)
DeepSeek V4 Pro 081387.0%SourceKimi K388.0%SourceKimi K3 leads this result
HLE w/o tools
DeepSeek V4 Pro 0813—Kimi K343.5%SourceNot directly comparable
Math
Multimodal
OfficeQA Pro
DeepSeek V4 Pro 0813—Kimi K363.3%SourceNot directly comparable
MMMU-Pro
DeepSeek V4 Pro 0813—Kimi K381.6%SourceNot directly comparable
MMMU-Pro w/ Python
DeepSeek V4 Pro 0813—Kimi K383.4%SourceNot directly comparable
CharXiv w/o tools
DeepSeek V4 Pro 0813—Kimi K384.8%SourceNot directly comparable
CharXiv
DeepSeek V4 Pro 0813—Kimi K391.3%SourceNot directly comparable
MathVision
DeepSeek V4 Pro 0813—Kimi K394.3%SourceNot directly comparable
MathVision w/ Python
DeepSeek V4 Pro 0813—Kimi K397.8%SourceNot directly comparable
BabyVision w/ Python
DeepSeek V4 Pro 0813—Kimi K385.7%SourceNot directly comparable
ZeroBench
DeepSeek V4 Pro 0813—Kimi K323.0%SourceNot directly comparable
ZeroBench w/ Python
DeepSeek V4 Pro 0813—Kimi K341.0%SourceNot directly comparable
WorldVQA ForceAnswer
DeepSeek V4 Pro 0813—Kimi K351.0%SourceNot directly comparable
OmniDocBench
DeepSeek V4 Pro 0813—Kimi K391.1%SourceNot directly comparable
PerceptionBench
DeepSeek V4 Pro 0813—Kimi K358.5%SourceNot directly comparable
Frequently asked questions
Which is better, DeepSeek V4 Pro 0813 or Kimi K3?
Kimi K3 has the higher public score estimate, 74.87 versus 66.35, but the 90% score intervals overlap. The higher estimate is not a decisive winner because the uncertainty ranges overlap.
Which is better for coding, DeepSeek V4 Pro 0813 or Kimi K3?
Kimi K3 leads the public coding lane, 68 to 52.2, with Supported evidence for both models, although the 90% intervals overlap.
Which is better for agentic tasks, DeepSeek V4 Pro 0813 or Kimi K3?
Kimi K3 leads the public agentic tasks lane, 71.9 to 56.1, with Supported evidence for both models and non-overlapping 90% intervals.
Which costs less, DeepSeek V4 Pro 0813 or Kimi K3?
For the stated presets, chat costs $0.00087 on DeepSeek V4 Pro 0813 and $0.0105 on Kimi K3; repository review costs $0.02436 and $0.195; the cache-heavy agent loop costs $0.01812 and $0.27. Costs use the listed standard API rates.
Which has the larger context window, DeepSeek V4 Pro 0813 or Kimi K3?
Kimi K3 has the larger documented context window: 1.05M, compared with 1M.
Related comparisons
Last updated September 4, 2026
Watch DeepSeek V4 Pro 0813 vs Kimi K3
One weekly email when material rank, price, or benchmark evidence changes make this matchup worth revisiting.
Read a sample issueJoin 2,000+ readers.