Agentic
Like-for-like- Claude Opus 5.5
- 82.7
- Supported · #1/156
- Kimi K3
- 72.2
- Supported · #5/156
- Basis
- BenchAlign lane · 11 vs 12 public rows
- Reading
- Claude Opus 5.5 leads · intervals overlap
Claude Opus 5.5 vs Kimi K3
Decision reading
Claude Opus 5.5 has the higher public score estimate, 81.03 versus 74.08, but the 90% score intervals overlap. Treat that as a lead, not a settled winner.
8 results are shared. Category rows resting on Estimated evidence or different benchmark sets are marked directional and do not name a winner.
Updated September 22, 2026. Rank says Claude Opus 5.5 is ahead. Price, access, and your workload can each overturn that. Public scores include evidence status and uncertainty.
Both of these models will change. Get the price, version and retirement notices for the pair, each with its source. Follow model changes
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.
Agentic work
Tool use, computer use, and multi-step task completion
Claude Opus 5.5
Claude Opus 5.5 leads on the public agentic lane, 82.7 to 72.2, with Supported evidence for both models, although the 90% intervals overlap.
Confidence: limited
Long documents
Prompts that approach the documented context limit
Kimi K3
Kimi K3 has the larger documented context window.
Confidence: documented
Chat turn cost
1K fresh input + 500 output tokens
Kimi K3
Kimi K3 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
Cache-heavy agent loop cost
200K cached + 20K fresh input + 10K output tokens
Kimi K3
Kimi K3 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
Kimi K3
Kimi K3 has the lower estimated token cost for this stated workload. Costs use the listed standard API rates.
Confidence: listed-rates
Coding work
Code generation, repair, and software-engineering tasks
Not enough matched evidence
Claude Opus 5.5 is scored on Estimated evidence for coding, so the reading is directional rather than like-for-like.
Confidence: limited
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.
76.0Claude Opus 5.566.8Kimi K3
Directional only · BenchAlign
Claude Opus 5.5 scores higher, but at least one score rests on Estimated evidence or a different benchmark set. Directional only, no winner.
Coding scores combine specific tasks and setups. Match the editor, harness, and effort to your workflow.
What is actually comparable
Shared results can support a head-to-head reading. Results present for only one model describe coverage, not superiority.
40 Claude Opus 5.5 only
8 Shared
36 Kimi K3 only
- Shared results
- 8
- Claude Opus 5.5 only
- 40
- Kimi K3 only
- 36
- Like-for-like categories
- 1 / 8
4 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
Directional only- Claude Opus 5.5
- 76.0
- Estimated · #4/159
- Kimi K3
- 66.8
- Supported · #11/159
- Basis
- BenchAlign lane · 9 vs 13 public rows
- Reading
- Directional only
Reasoning
Directional only- Claude Opus 5.5
- 78.5
- #4/19
- Kimi K3
- 78.5
- #3/19
- Basis
- Provisional lane · 0 vs 0 weighted rows
- Reading
- Directional only
Multimodal
Directional only- Claude Opus 5.5
- 88.8
- #3/50
- Kimi K3
- 89.4
- #1/50
- Basis
- Provisional lane · 1 vs 3 weighted rows
- Reading
- Directional only
Knowledge
Directional only- Claude Opus 5.5
- 81.5
- Estimated · #3/187
- Kimi K3
- 71.3
- Supported · #9/187
- Basis
- BenchAlign lane · 17 vs 6 public rows
- Reading
- Directional only
Multilingual
Not comparable- Claude Opus 5.5
- Not ranked
- Kimi K3
- Not ranked
- Basis
- Provisional lane · 0 vs 0 weighted rows
- Reading
- Not comparable
Instruction following
Not comparable- Claude Opus 5.5
- Not ranked
- Kimi K3
- Not ranked
- Basis
- Provisional lane · 0 vs 0 weighted rows
- Reading
- Not comparable
Math
Not comparable- Claude Opus 5.5
- Not ranked
- Kimi K3
- Not ranked
- Basis
- Provisional lane · 0 vs 0 weighted rows
- Reading
- Not comparable
| Category | Claude Opus 5.5 | Kimi K3 | Basis | Reading |
|---|---|---|---|---|
| Agentic | 82.7Supported · #1/156 | 72.2Supported · #5/156 | Like-for-likeBenchAlign lane · 11 vs 12 public rows | Claude Opus 5.5 leads · intervals overlap |
| Coding | 76.0Estimated · #4/159 | 66.8Supported · #11/159 | Directional onlyBenchAlign lane · 9 vs 13 public rows | Directional only |
| Reasoning | 78.5#4/19 | 78.5#3/19 | Directional onlyProvisional lane · 0 vs 0 weighted rows | Directional only |
| Multimodal | 88.8#3/50 | 89.4#1/50 | Directional onlyProvisional lane · 1 vs 3 weighted rows | Directional only |
| Knowledge | 81.5Estimated · #3/187 | 71.3Supported · #9/187 | Directional onlyBenchAlign lane · 17 vs 6 public rows | Directional only |
| 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) 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.
HLE w/o tools
Knowledge
HLE
Knowledge
AutomationBench
Agentic
OfficeQA 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
- Claude Opus 5.5
- $0.014
- Fits in one request
- Kimi K3
- $0.0105
- Fits in one request
Kimi K3 has the lower modeled cost
Costs use the listed standard API rates.
Repository review
50K fresh input + 3K output tokens
- Claude Opus 5.5
- $0.26
- Fits in one request
- Kimi K3
- $0.195
- Fits in one request
Kimi K3 has the lower modeled cost
Costs use the listed standard API rates.
Cache-heavy agent loop
200K cached + 20K fresh input + 10K output tokens
- Claude Opus 5.5
- $0.32
- Fits in one request
- Kimi K3
- $0.27
- Fits in one request
Kimi K3 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.
SpecificationClaude Opus 5.5Kimi K3
Context window
Maximum documented context; output-token limits may be lower.
Claude Opus 5.5
Kimi K3
1.05M
API model ID
Claude Opus 5.5
claude-opus-5-5
Anthropic Claude Opus 5.5 model documentationKimi K3
Not sourced
Cached-input rate
A missing cached-input rate falls back to the listed input rate only in the stated workload estimate.
Claude Opus 5.5
$0.2 per 1M cached input tokens
Claude Opus 5.5 model documentationKimi K3
$0.3 per 1M cached input tokens
Documented inputs
Claude Opus 5.5
Kimi K3
Not sourced
Documented outputs
Claude Opus 5.5
Kimi K3
Not sourced
Provider availability
Claude Opus 5.5
Generally Available · Claude API, Amazon Bedrock, Google Cloud, Microsoft Foundry, Claude Platform on AWS
Anthropic Claude Opus 5.5 model documentationKimi K3
Not sourced
Reasoning profile
Claude Opus 5.5
Reasoning
Kimi K3
Reasoning
Weight access
Claude Opus 5.5
Proprietary
Kimi K3
Pending
License
Claude Opus 5.5
Proprietary
Kimi K3
Pending
Release date
Claude Opus 5.5
2026-09-22
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
- Claude Opus 5.5 has the higher public score estimate, 81.03 versus 74.08, but the 90% score intervals overlap.
- Workload cost
- Repository review: $0.26 vs $0.195. Cache-heavy agent loop: $0.32 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 evidence84 rows
Agentic
Terminal-Bench 4.0
Claude Opus 5.566.40%SourceKimi K3—Not directly comparable
Terminal-Bench-Science 0.1
Claude Opus 5.558.7%SourceKimi K3—Not directly comparable
AutomationBench
Claude Opus 5.540.0%SourceKimi K330.8%SourceClaude Opus 5.5 leads this result
HLE w/ tools
Claude Opus 5.567.7%SourceKimi K3—Not directly comparable
OSWorld 2.0
Claude Opus 5.548.7%SourceKimi K3—Not directly comparable
LAB all-pass (Harvey held-out)
Claude Opus 5.58.3%SourceKimi K3—Not directly comparable
LAB criterion-pass (Harvey held-out)
Claude Opus 5.591.2%SourceKimi K3—Not directly comparable
Toolathlon-Verified
Claude Opus 5.577.8%SourceKimi K373.2%SourceClaude Opus 5.5 leads this result
Toolathlon Verified Pass@3
Claude Opus 5.582.4%SourceKimi K3—Not directly comparable
Toolathlon Verified Pass³
Claude Opus 5.572.2%SourceKimi K3—Not directly comparable
Toolathlon Verified avg. turns
Claude Opus 5.526.9 turnsSourceKimi K3—Not directly comparable
Terminal-Bench 2.0
Claude Opus 5.5—Kimi K388.3%SourceNot directly comparable
BrowseComp
Claude Opus 5.5—Kimi K391.2%SourceNot directly comparable
DeepSearchQA
Claude Opus 5.5—Kimi K395.0%SourceNot directly comparable
MCP Atlas
Claude Opus 5.5—Kimi K384.2%SourceNot directly comparable
JobBench
Claude Opus 5.5—Kimi K352.9%SourceNot directly comparable
APEX-Agents
Claude Opus 5.5—Kimi K337.6%SourceNot directly comparable
SpreadsheetBench 2
Claude Opus 5.5—Kimi K334.8%SourceNot directly comparable
DECK-Bench
Claude Opus 5.5—Kimi K373.5%SourceNot directly comparable
Terminal-Bench 2.1 (Vals)
Claude Opus 5.5—Kimi K380.9%SourceNot directly comparable
ApprenticeBench
Claude Opus 5.5—Kimi K318%SourceNot directly comparable
Coding
FrontierCode 1.1 Main
Claude Opus 5.554.4%SourceKimi K3—Not directly comparable
cursorBench40
Claude Opus 5.557.8%SourceKimi K3—Not directly comparable
SWE-bench Pro
Claude Opus 5.589.9%SourceKimi K3—Not directly comparable
SWE Multilingual
Claude Opus 5.593.9%SourceKimi K3—Not directly comparable
SWE Multimodal
Claude Opus 5.561.4%SourceKimi K3—Not directly comparable
DeepSWE
Claude Opus 5.574.2%SourceKimi K367.5%SourceClaude Opus 5.5 leads this result
FrontierCode 1.1 Extended
Claude Opus 5.563.6%SourceKimi K3—Not directly comparable
FrontierSWE v2
Claude Opus 5.562.3%SourceKimi K325.9%SourceClaude Opus 5.5 leads this result
ProgramBench
Claude Opus 5.591.2%SourceKimi K377.8%SourceClaude Opus 5.5 leads this result
cursorBench32
Claude Opus 5.5—Kimi K360.8%SourceNot directly comparable
FrontierSWE
Claude Opus 5.5—Kimi K381.2%SourceNot directly comparable
Kimi Code Bench v2
Claude Opus 5.5—Kimi K372.9%SourceNot directly comparable
sweMarathon
Claude Opus 5.5—Kimi K342%SourceNot directly comparable
PostTrain Bench
Claude Opus 5.5—Kimi K336.6%SourceNot directly comparable
MLS-Bench Lite
Claude Opus 5.5—Kimi K348.3%SourceNot directly comparable
VulcanBench v3
Claude Opus 5.5—Kimi K373.7%SourceNot directly comparable
OpenHarmony Bench
Claude Opus 5.5—Kimi K357.3%SourceNot directly comparable
LiveCodeBench (Vals)
Claude Opus 5.5—Kimi K387.2%SourceNot directly comparable
SWE-bench (Vals)
Claude Opus 5.5—Kimi K393.4%SourceNot directly comparable
Multimodal
Chartography (tools)
Claude Opus 5.589.0%SourceKimi K3—Not directly comparable
Chartography (no tools)
Claude Opus 5.564.4%SourceKimi K3—Not directly comparable
BenchCAD Vision2Code (no tools)
Claude Opus 5.50.730SourceKimi K3—Not directly comparable
BenchCAD Vision2Code (tools)
Claude Opus 5.50.962SourceKimi K3—Not directly comparable
Biomedical image analysis
Claude Opus 5.571.4%SourceKimi K3—Not directly comparable
OfficeQA
Claude Opus 5.578.9%SourceKimi K3—Not directly comparable
OfficeQA Pro
Claude Opus 5.567.7%SourceKimi K363.3%SourceClaude Opus 5.5 leads this result
MMMU-Pro
Claude Opus 5.5—Kimi K381.6%SourceNot directly comparable
MMMU-Pro w/ Python
Claude Opus 5.5—Kimi K383.4%SourceNot directly comparable
CharXiv w/o tools
Claude Opus 5.5—Kimi K384.8%SourceNot directly comparable
CharXiv
Claude Opus 5.5—Kimi K391.3%SourceNot directly comparable
MathVision
Claude Opus 5.5—Kimi K394.3%SourceNot directly comparable
MathVision w/ Python
Claude Opus 5.5—Kimi K397.8%SourceNot directly comparable
BabyVision w/ Python
Claude Opus 5.5—Kimi K385.7%SourceNot directly comparable
ZeroBench
Claude Opus 5.5—Kimi K323.0%SourceNot directly comparable
ZeroBench w/ Python
Claude Opus 5.5—Kimi K341.0%SourceNot directly comparable
WorldVQA ForceAnswer
Claude Opus 5.5—Kimi K351.0%SourceNot directly comparable
OmniDocBench
Claude Opus 5.5—Kimi K391.1%SourceNot directly comparable
PerceptionBench
Claude Opus 5.5—Kimi K358.5%SourceNot directly comparable
Knowledge
HLE
Claude Opus 5.567.7%SourceKimi K356%SourceClaude Opus 5.5 leads this result
HLE w/o tools
Claude Opus 5.564.4%SourceKimi K343.5%SourceClaude Opus 5.5 leads this result
HealthBench (raw)
Claude Opus 5.568.1%SourceKimi K3—Not directly comparable
HealthBench (length-adjusted)
Claude Opus 5.560.6%SourceKimi K3—Not directly comparable
HealthBench Professional
Claude Opus 5.565.6%SourceKimi K3—Not directly comparable
HealthBench Professional (raw)
Claude Opus 5.577.1%SourceKimi K3—Not directly comparable
BioMysteryBench (human-solvable)
Claude Opus 5.589.3%SourceKimi K3—Not directly comparable
BioMysteryBench (human-difficult)
Claude Opus 5.550.0%SourceKimi K3—Not directly comparable
SpatialBench Verified
Claude Opus 5.572.0%SourceKimi K3—Not directly comparable
SingleCellBench
Claude Opus 5.561.2%SourceKimi K3—Not directly comparable
Morphology-to-molecule matching
Claude Opus 5.534.0%SourceKimi K3—Not directly comparable
Medicinal chemistry
Claude Opus 5.563.5%SourceKimi K3—Not directly comparable
Protein Design
Claude Opus 5.560.2%SourceKimi K3—Not directly comparable
Protein Design library ranking
Claude Opus 5.556.0%SourceKimi K3—Not directly comparable
De novo protein-binder design
Claude Opus 5.582.6%SourceKimi K3—Not directly comparable
Protocols (troubleshooting)
Claude Opus 5.573.7%SourceKimi K3—Not directly comparable
Protocols (understanding)
Claude Opus 5.569.0%SourceKimi K3—Not directly comparable
GPQA
Claude Opus 5.5—Kimi K393.5%SourceNot directly comparable
GPQA-D
Claude Opus 5.5—Kimi K393.5%SourceNot directly comparable
GPQA Diamond (Vals)
Claude Opus 5.5—Kimi K392.9%SourceNot directly comparable
MMLU-Pro (Vals)
Claude Opus 5.5—Kimi K388.0%SourceNot directly comparable
Multilingual
Math
Questions
Which is better, Claude Opus 5.5 or Kimi K3?
Claude Opus 5.5 has the higher public score estimate, 81.03 versus 74.08, but the 90% score intervals overlap. The higher estimate is not a decisive winner because the uncertainty ranges overlap.
Which is better for coding, Claude Opus 5.5 or Kimi K3?
Claude Opus 5.5 scores higher for coding on the public lane, 76 to 66.8. Claude Opus 5.5 is scored on Estimated evidence for coding, so the reading is directional rather than like-for-like. BenchLM does not name a winner for coding from a directional reading; read the shared benchmark rows directly and test the models on the same task set.
Which is better for agentic tasks, Claude Opus 5.5 or Kimi K3?
Claude Opus 5.5 leads the public agentic tasks lane, 82.7 to 72.2, with Supported evidence for both models, although the 90% intervals overlap.
Which costs less, Claude Opus 5.5 or Kimi K3?
For the stated presets, chat costs $0.014 on Claude Opus 5.5 and $0.0105 on Kimi K3; repository review costs $0.26 and $0.195; the cache-heavy agent loop costs $0.32 and $0.27. Costs use the listed standard API rates.
Which has the larger context window, Claude Opus 5.5 or Kimi K3?
Kimi K3 has the larger documented context window: 1.05M, compared with 1M.
Related comparisons
Last updated September 22, 2026
Watch Claude Opus 5.5 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.