Coding work
Code generation, repair, and software-engineering tasks
Kimi K3
Kimi K3 leads on the public coding lane, 62.6 to 58.1, with Supported evidence for both models, although the 90% intervals overlap.
Confidence: limited
Grok 4.5 vs Kimi K3
Updated September 27, 2026. Rank says Kimi K3 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
Decision reading
Kimi K3 has the higher public score estimate, 71.87 versus 65.27, but the 90% score intervals overlap. Treat that as a lead, not a settled winner. 9 results are shared. Category rows resting on Estimated evidence or different benchmark sets are marked directional and do not name a winner.
- Shared results
- 9
- Grok 4.5 only
- 6
- Kimi K3 only
- 37
- Like-for-like categories
- 3 / 8
Supported: Grok 4.5 and 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.
Agentic work
Tool use, computer use, and multi-step task completion
Kimi K3
Kimi K3 leads on the public agentic lane, 70 to 57.9, with Supported evidence for both models and non-overlapping 90% intervals.
Confidence: strongerLong 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
Grok 4.5
Grok 4.5 has the lower estimated token cost for this stated workload. Costs use the listed standard API rates.
Confidence: listed-ratesCache-heavy agent loop cost
200K cached + 20K fresh input + 10K output tokens
Grok 4.5
Grok 4.5 has the lower estimated token cost for this stated workload. Costs use the listed standard API rates.
Confidence: listed-ratesRepository review cost
50K fresh input + 3K output tokens
Grok 4.5
Grok 4.5 has the lower estimated token cost for this stated workload. Costs use the listed standard API rates.
Confidence: listed-rates
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.
58.1Grok 4.562.6Kimi K3
Like-for-like · BenchAlign v5.7
Kimi K3 leads the like-for-like coding row, although the 90% intervals overlap.
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.
6 Grok 4.5 only
9 Shared
37 Kimi K3 only
1 category rests on Estimated evidence or different evidence sets. Those rows remain visible for coverage context but do not name a winner.
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.
ARC-AGI-2Reasoning
Normalized gap 7.8Grok 4.5:52.6%Kimi K3:60.4%cursorBench32Coding
Normalized gap 5.9Grok 4.5:66.7%Kimi K3:60.8%MMLU-Pro (Vals)Knowledge
Normalized gap 1.2Grok 4.5:89.2%Kimi K3:88.0%LiveCodeBench (Vals)Coding
Normalized gap 0.2Grok 4.5:87.4%Kimi K3:87.2%
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 v5.7 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.
Agentic
Like-for-like- Grok 4.5
- 57.9
- Supported · #21/105
- Kimi K3
- 70.0
- Supported · #6/105
- Basis
- BenchAlign v5.7 lane · 4 vs 12 public rows
- Reading
- Kimi K3 leads
Coding
Like-for-like- Grok 4.5
- 58.1
- Supported · #19/135
- Kimi K3
- 62.6
- Supported · #13/135
- Basis
- BenchAlign v5.7 lane · 7 vs 13 public rows
- Reading
- Kimi K3 leads · intervals overlap
Knowledge
Like-for-like- Grok 4.5
- 66.8
- Supported · #17/158
- Kimi K3
- 68.5
- Supported · #14/158
- Basis
- BenchAlign v5.7 lane · 2 vs 6 public rows
- Reading
- Kimi K3 leads · intervals overlap
Reasoning
Directional only- Grok 4.5
- 50.3
- #19/19
- Kimi K3
- 65.4
- #11/19
- Basis
- Provisional lane · 2 vs 1 weighted rows
- Reading
- Directional only
Multimodal
Not comparable- Grok 4.5
- 78.5
- Unranked · 1 rankable row
- Kimi K3
- 89.4
- #1/50
- Basis
- Provisional lane · 0 vs 3 weighted rows
- Reading
- Not comparable
Multilingual
Not comparable- Grok 4.5
- Not ranked
- Kimi K3
- Not ranked
- Basis
- Provisional lane · 0 vs 0 weighted rows
- Reading
- Not comparable
Instruction following
Not comparable- Grok 4.5
- Not ranked
- Kimi K3
- Not ranked
- Basis
- Provisional lane · 0 vs 0 weighted rows
- Reading
- Not comparable
Math
Not comparable- Grok 4.5
- Not ranked
- Kimi K3
- Not ranked
- Basis
- Provisional lane · 0 vs 0 weighted rows
- Reading
- Not comparable
| Category | Grok 4.5 | Kimi K3 | Basis | Reading |
|---|---|---|---|---|
| Agentic | 57.9Supported · #21/105 | 70.0Supported · #6/105 | Like-for-likeBenchAlign v5.7 lane · 4 vs 12 public rows | Kimi K3 leads |
| Coding | 58.1Supported · #19/135 | 62.6Supported · #13/135 | Like-for-likeBenchAlign v5.7 lane · 7 vs 13 public rows | Kimi K3 leads · intervals overlap |
| Knowledge | 66.8Supported · #17/158 | 68.5Supported · #14/158 | Like-for-likeBenchAlign v5.7 lane · 2 vs 6 public rows | Kimi K3 leads · intervals overlap |
| Reasoning | 50.3#19/19 | 65.4#11/19 | Directional onlyProvisional lane · 2 vs 1 weighted rows | Directional only |
| Multimodal | 78.5Unranked · 1 rankable row | 89.4#1/50 | Not comparableProvisional lane · 0 vs 3 weighted 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 v5.7) 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
- Grok 4.5
- $0.005
- Fits in one request
- Kimi K3
- $0.0105
- Fits in one request
Grok 4.5 has the lower modeled cost
Costs use the listed standard API rates.
Repository review
50K fresh input + 3K output tokens
- Grok 4.5
- $0.118
- Fits in one request
- Kimi K3
- $0.195
- Fits in one request
Grok 4.5 has the lower modeled cost
Costs use the listed standard API rates.
Cache-heavy agent loop
200K cached + 20K fresh input + 10K output tokens
- Grok 4.5
- $0.16
- Fits in one request
- Kimi K3
- $0.27
- Fits in one request
Grok 4.5 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.
SpecificationGrok 4.5Kimi K3
Context window
Maximum documented context; output-token limits may be lower.
Grok 4.5
500K
Kimi K3
1.05M
API model ID
Grok 4.5
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.
Grok 4.5
$0.3 per 1M cached input tokens
Kimi K3
$0.3 per 1M cached input tokens
Documented inputs
Grok 4.5
Not sourced
Kimi K3
Not sourced
Documented outputs
Grok 4.5
Not sourced
Kimi K3
Not sourced
Provider availability
Grok 4.5
Not sourced
Kimi K3
Not sourced
Reasoning profile
Grok 4.5
Reasoning
Kimi K3
Reasoning
Weight access
Grok 4.5
Proprietary
Kimi K3
Pending
License
Grok 4.5
Proprietary
Kimi K3
Pending
Release date
Grok 4.5
2026-07-08
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, 71.87 versus 65.27, but the 90% score intervals overlap.
- Workload cost
- Repository review: $0.118 vs $0.195. Cache-heavy agent loop: $0.16 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.
Questions
Which is better, Grok 4.5 or Kimi K3?
Kimi K3 has the higher public score estimate, 71.87 versus 65.27, but the 90% score intervals overlap. The higher estimate is not a decisive winner because the uncertainty ranges overlap.
Which is better for coding, Grok 4.5 or Kimi K3?
Kimi K3 leads the public coding lane, 62.6 to 58.1, with Supported evidence for both models, although the 90% intervals overlap.
Which is better for agentic tasks, Grok 4.5 or Kimi K3?
Kimi K3 leads the public agentic tasks lane, 70 to 57.9, with Supported evidence for both models and non-overlapping 90% intervals.
Which costs less, Grok 4.5 or Kimi K3?
For the stated presets, chat costs $0.005 on Grok 4.5 and $0.0105 on Kimi K3; repository review costs $0.118 and $0.195; the cache-heavy agent loop costs $0.16 and $0.27. Costs use the listed standard API rates.
Which has the larger context window, Grok 4.5 or Kimi K3?
Kimi K3 has the larger documented context window: 1.05M, compared with 500K.
Benchmark evidence
The full public result ledger is available for audit without forcing a wide desktop table onto a phone.
Browse raw public benchmark evidence52 rows
Agentic
Terminal-Bench 3.0
Grok 4.515.7%SourceKimi K3—Not directly comparable
Terminal-Bench 2.1
Grok 4.583.3%SourceKimi K388.3%SourceKimi K3 leads this result
deepSwe
Grok 4.553%SourceKimi K3—Not directly comparable
Terminal-Bench 2.1 (Vals)
Grok 4.567.8%SourceKimi K380.9%SourceKimi K3 leads this result
BrowseComp
Grok 4.5—Kimi K391.2%SourceNot directly comparable
DeepSearchQA
Grok 4.5—Kimi K395.0%SourceNot directly comparable
Toolathlon-Verified
Grok 4.5—Kimi K373.2%SourceNot directly comparable
MCP Atlas
Grok 4.5—Kimi K384.2%SourceNot directly comparable
AutomationBench
Grok 4.5—Kimi K330.8%SourceNot directly comparable
JobBench
Grok 4.5—Kimi K352.9%SourceNot directly comparable
APEX-Agents
Grok 4.5—Kimi K337.6%SourceNot directly comparable
SpreadsheetBench 2
Grok 4.5—Kimi K334.8%SourceNot directly comparable
DECK-Bench
Grok 4.5—Kimi K373.5%SourceNot directly comparable
ApprenticeBench
Grok 4.5—Kimi K318%SourceNot directly comparable
Coding
SWE-bench Pro
Grok 4.564.7%SourceKimi K3—Not directly comparable
SWE Multilingual
Grok 4.578%SourceKimi K3—Not directly comparable
Terminal-Bench 2.1
Grok 4.583.3%SourceKimi K3—Not directly comparable
cursorBench32
Shared sourceGrok 4.566.7%Kimi K360.8%Grok 4.5 leads this result
VulcanBench v3
Grok 4.589.9%SourceKimi K373.7%SourceGrok 4.5 leads this result
LiveCodeBench (Vals)
Grok 4.587.4%SourceKimi K387.2%SourceGrok 4.5 leads this result
SWE-bench (Vals)
Grok 4.586.6%SourceKimi K393.4%SourceKimi K3 leads this result
DeepSWE
Grok 4.5—Kimi K367.5%SourceNot directly comparable
FrontierSWE
Grok 4.5—Kimi K381.2%SourceNot directly comparable
ProgramBench
Grok 4.5—Kimi K377.8%SourceNot directly comparable
Kimi Code Bench v2
Grok 4.5—Kimi K372.9%SourceNot directly comparable
sweMarathon
Grok 4.5—Kimi K342%SourceNot directly comparable
PostTrain Bench
Grok 4.5—Kimi K336.6%SourceNot directly comparable
MLS-Bench Lite
Grok 4.5—Kimi K348.3%SourceNot directly comparable
OpenHarmony Bench
Grok 4.5—Kimi K357.3%SourceNot directly comparable
FrontierSWE v2
Grok 4.5—Kimi K325.9%SourceNot directly comparable
Reasoning
Multimodal
OfficeQA Pro
Grok 4.5—Kimi K363.3%SourceNot directly comparable
MMMU-Pro
Grok 4.5—Kimi K381.6%SourceNot directly comparable
MMMU-Pro w/ Python
Grok 4.5—Kimi K383.4%SourceNot directly comparable
CharXiv w/o tools
Grok 4.5—Kimi K384.8%SourceNot directly comparable
CharXiv
Grok 4.5—Kimi K391.3%SourceNot directly comparable
MathVision
Grok 4.5—Kimi K394.3%SourceNot directly comparable
MathVision w/ Python
Grok 4.5—Kimi K397.8%SourceNot directly comparable
BabyVision w/ Python
Grok 4.5—Kimi K385.7%SourceNot directly comparable
ZeroBench
Grok 4.5—Kimi K323.0%SourceNot directly comparable
ZeroBench w/ Python
Grok 4.5—Kimi K341.0%SourceNot directly comparable
WorldVQA ForceAnswer
Grok 4.5—Kimi K351.0%SourceNot directly comparable
OmniDocBench
Grok 4.5—Kimi K391.1%SourceNot directly comparable
PerceptionBench
Grok 4.5—Kimi K358.5%SourceNot directly comparable
Knowledge
GPQA Diamond (Vals)
Grok 4.592.9%SourceKimi K392.9%SourceTie
MMLU-Pro (Vals)
Grok 4.589.2%SourceKimi K388.0%SourceGrok 4.5 leads this result
GPQA
Grok 4.5—Kimi K393.5%SourceNot directly comparable
GPQA-D
Grok 4.5—Kimi K393.5%SourceNot directly comparable
HLE
Grok 4.5—Kimi K356%SourceNot directly comparable
HLE w/o tools
Grok 4.5—Kimi K343.5%SourceNot directly comparable
52 public results · 9 shared
Watch Grok 4.5 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.
Last updated September 27, 2026