Agentic
Like-for-like- Claude Fable 5
- 74.8
- Supported · #3/151
- Kimi K3
- 71.8
- Supported · #4/151
- Basis
- BenchAlign lane · 4 vs 11 public rows
- Reading
- Claude Fable 5 leads · intervals overlap
Claude Fable 5 vs Kimi K3
Updated September 4, 2026. Public scores include evidence status and uncertainty. They are not guarantees for a specific workload.
Decision reading
Claude Fable 5 has the higher public score, 80.91 versus 74.87, and the 90% score intervals do not overlap.
10 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
Claude Fable 5
Claude Fable 5 leads on the public coding lane, 76.9 to 68, with Supported evidence for both models, although the 90% intervals overlap.
Confidence: limited
Agentic work
Tool use, computer use, and multi-step task completion
Claude Fable 5
Claude Fable 5 leads on the public agentic lane, 74.8 to 71.8, 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
Show secondary and unsupported calls
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
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
What is actually comparable
Shared results can support a head-to-head reading. Results present for only one model describe coverage, not superiority.
8 Claude Fable 5 only
10 Shared
33 Kimi K3 only
- Shared results
- 10
- Claude Fable 5 only
- 8
- Kimi K3 only
- 33
- Like-for-like categories
- 3 / 8
1 category rests 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- Claude Fable 5
- 76.9
- Supported · #2/183
- Kimi K3
- 68.0
- Supported · #7/183
- Basis
- BenchAlign lane · 10 vs 13 public rows
- Reading
- Claude Fable 5 leads · intervals overlap
Knowledge
Like-for-like- Claude Fable 5
- 83.6
- Supported · #2/181
- Kimi K3
- 72.3
- Supported · #8/181
- Basis
- BenchAlign lane · 2 vs 6 public rows
- Reading
- Claude Fable 5 leads · intervals overlap
Reasoning
Directional only- Claude Fable 5
- 76.2
- #11/22
- Kimi K3
- 78.5
- #3/22
- Basis
- Provisional lane · 0 vs 0 weighted rows
- Reading
- Directional only
Math
Not comparable- Claude Fable 5
- Not ranked
- Kimi K3
- Not ranked
- Basis
- Provisional lane · 0 vs 0 weighted rows
- Reading
- Not comparable
Multilingual
Not comparable- Claude Fable 5
- Not ranked
- Kimi K3
- Not ranked
- Basis
- Provisional lane · 0 vs 0 weighted rows
- Reading
- Not comparable
Multimodal
Not comparable- Claude Fable 5
- 62.3
- Unranked · 2 rankable rows
- Kimi K3
- 88.8
- #1/48
- Basis
- Provisional lane · 1 vs 3 weighted rows
- Reading
- Not comparable
Instruction following
Not comparable- Claude Fable 5
- 78.3
- #54/120
- Kimi K3
- Not ranked
- Basis
- Provisional lane · 0 vs 0 weighted rows
- Reading
- Not comparable
| Category | Claude Fable 5 | Kimi K3 | Basis | Reading |
|---|---|---|---|---|
| Agentic | 74.8Supported · #3/151 | 71.8Supported · #4/151 | Like-for-likeBenchAlign lane · 4 vs 11 public rows | Claude Fable 5 leads · intervals overlap |
| Coding | 76.9Supported · #2/183 | 68.0Supported · #7/183 | Like-for-likeBenchAlign lane · 10 vs 13 public rows | Claude Fable 5 leads · intervals overlap |
| Knowledge | 83.6Supported · #2/181 | 72.3Supported · #8/181 | Like-for-likeBenchAlign lane · 2 vs 6 public rows | Claude Fable 5 leads · intervals overlap |
| Reasoning | 76.2#11/22 | 78.5#3/22 | Directional onlyProvisional lane · 0 vs 0 weighted rows | Directional only |
| Math | Not ranked | Not ranked | Not comparableProvisional lane · 0 vs 0 weighted rows | Not comparable |
| Multilingual | Not ranked | Not ranked | Not comparableProvisional lane · 0 vs 0 weighted rows | Not comparable |
| Multimodal | 62.3Unranked · 2 rankable rows | 88.8#1/48 | Not comparableProvisional lane · 1 vs 3 weighted rows | Not comparable |
| Instruction following | 78.3#54/120 | 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.
cursorBench32
Coding
OfficeQA Pro
Multimodal
Terminal-Bench 2.0
Agentic
MMLU-Pro (Vals)
Knowledge
LiveCodeBench (Vals)
Coding
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 Fable 5
- $0.035
- 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 Fable 5
- $0.65
- 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 Fable 5
- $0.9
- 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 Fable 5Kimi K3
Context window
Maximum documented context; output-token limits may be lower.
Claude Fable 5
Kimi K3
1.05M
API model ID
Claude Fable 5
claude-fable-5
Anthropic model overviewKimi 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 Fable 5
$1 per 1M cached input tokens
Claude API pricingKimi K3
$0.3 per 1M cached input tokens
Documented inputs
Claude Fable 5
text, image
Anthropic model overviewKimi K3
Not sourced
Documented outputs
Claude Fable 5
Kimi K3
Not sourced
Provider availability
Claude Fable 5
Generally Available · Claude API
Anthropic model overviewKimi K3
Not sourced
Reasoning profile
Claude Fable 5
Reasoning
Kimi K3
Reasoning
Weight access
Claude Fable 5
Proprietary
Kimi K3
Pending
License
Claude Fable 5
Proprietary
Kimi K3
Pending
Release date
Claude Fable 5
2026-06-09
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 Fable 5 has the higher public score, 80.91 versus 74.87, and the 90% score intervals do not overlap.
- Workload cost
- Repository review: $0.65 vs $0.195. Cache-heavy agent loop: $0.9 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 evidence51 rows
Agentic
Terminal-Bench 3.0
Claude Fable 534.0%SourceKimi K3—Not directly comparable
Terminal-Bench 2.0
Claude Fable 584.3%SourceKimi K388.3%SourceKimi K3 leads this result
OSWorld-Verified
Claude Fable 585%SourceKimi K3—Not directly comparable
Terminal-Bench 2.1 (Vals)
Claude Fable 580.5%SourceKimi K380.9%SourceKimi K3 leads this result
BrowseComp
Claude Fable 5—Kimi K391.2%SourceNot directly comparable
DeepSearchQA
Claude Fable 5—Kimi K395.0%SourceNot directly comparable
Toolathlon-Verified
Claude Fable 5—Kimi K373.2%SourceNot directly comparable
MCP Atlas
Claude Fable 5—Kimi K384.2%SourceNot directly comparable
AutomationBench
Claude Fable 5—Kimi K330.8%SourceNot directly comparable
JobBench
Claude Fable 5—Kimi K352.9%SourceNot directly comparable
APEX-Agents
Claude Fable 5—Kimi K337.6%SourceNot directly comparable
SpreadsheetBench 2
Claude Fable 5—Kimi K334.8%SourceNot directly comparable
DECK-Bench
Claude Fable 5—Kimi K373.5%SourceNot directly comparable
Coding
SWE-bench Verified
Claude Fable 595%SourceKimi K3—Not directly comparable
SWE-bench Pro
Claude Fable 580%SourceKimi K3—Not directly comparable
FrontierSWE v2
Claude Fable 548.0%SourceKimi K325.9%SourceClaude Fable 5 leads this result
FrontierCode 1.1 Main
Claude Fable 553.5%SourceKimi K3—Not directly comparable
Terminal-Bench 2.0
Claude Fable 584.3%SourceKimi K3—Not directly comparable
cursorBench31
Claude Fable 570.6%SourceKimi K3—Not directly comparable
cursorBench32
Shared sourceClaude Fable 570.5%Kimi K360.8%Claude Fable 5 leads this result
VulcanBench v3
Claude Fable 589.5%SourceKimi K373.7%SourceClaude Fable 5 leads this result
LiveCodeBench (Vals)
Claude Fable 589.8%SourceKimi K387.2%SourceClaude Fable 5 leads this result
SWE-bench (Vals)
Claude Fable 595.0%SourceKimi K393.4%SourceClaude Fable 5 leads this result
deepSwe
Claude Fable 5—Kimi K367.5%SourceNot directly comparable
FrontierSWE
Claude Fable 5—Kimi K381.2%SourceNot directly comparable
ProgramBench
Claude Fable 5—Kimi K377.8%SourceNot directly comparable
Kimi Code Bench v2
Claude Fable 5—Kimi K372.9%SourceNot directly comparable
sweMarathon
Claude Fable 5—Kimi K342%SourceNot directly comparable
PostTrain Bench
Claude Fable 5—Kimi K336.6%SourceNot directly comparable
MLS-Bench Lite
Claude Fable 5—Kimi K348.3%SourceNot directly comparable
OpenHarmony Bench
Claude Fable 5—Kimi K357.3%SourceNot directly comparable
Knowledge
GPQA Diamond (Vals)
Claude Fable 593.2%SourceKimi K392.9%SourceClaude Fable 5 leads this result
MMLU-Pro (Vals)
Claude Fable 591.5%SourceKimi K388.0%SourceClaude Fable 5 leads this result
GPQA
Claude Fable 5—Kimi K393.5%SourceNot directly comparable
GPQA-D
Claude Fable 5—Kimi K393.5%SourceNot directly comparable
HLE
Claude Fable 5—Kimi K356%SourceNot directly comparable
HLE w/o tools
Claude Fable 5—Kimi K343.5%SourceNot directly comparable
Multimodal
Blueprint-Bench 2
Claude Fable 538.6%SourceKimi K3—Not directly comparable
OfficeQA Pro
Claude Fable 557.9%SourceKimi K363.3%SourceKimi K3 leads this result
MMMU-Pro
Claude Fable 5—Kimi K381.6%SourceNot directly comparable
MMMU-Pro w/ Python
Claude Fable 5—Kimi K383.4%SourceNot directly comparable
CharXiv w/o tools
Claude Fable 5—Kimi K384.8%SourceNot directly comparable
CharXiv
Claude Fable 5—Kimi K391.3%SourceNot directly comparable
MathVision
Claude Fable 5—Kimi K394.3%SourceNot directly comparable
MathVision w/ Python
Claude Fable 5—Kimi K397.8%SourceNot directly comparable
BabyVision w/ Python
Claude Fable 5—Kimi K385.7%SourceNot directly comparable
ZeroBench
Claude Fable 5—Kimi K323.0%SourceNot directly comparable
ZeroBench w/ Python
Claude Fable 5—Kimi K341.0%SourceNot directly comparable
WorldVQA ForceAnswer
Claude Fable 5—Kimi K351.0%SourceNot directly comparable
OmniDocBench
Claude Fable 5—Kimi K391.1%SourceNot directly comparable
PerceptionBench
Claude Fable 5—Kimi K358.5%SourceNot directly comparable
Frequently asked questions
Which is better, Claude Fable 5 or Kimi K3?
Claude Fable 5 has the higher public score, 80.91 versus 74.87, and the 90% score intervals do not overlap. That is the clearest overall quality signal in the current public data.
Which is better for coding, Claude Fable 5 or Kimi K3?
Claude Fable 5 leads the public coding lane, 76.9 to 68, with Supported evidence for both models, although the 90% intervals overlap.
Which is better for agentic tasks, Claude Fable 5 or Kimi K3?
Claude Fable 5 leads the public agentic tasks lane, 74.8 to 71.8, with Supported evidence for both models, although the 90% intervals overlap.
Which costs less, Claude Fable 5 or Kimi K3?
For the stated presets, chat costs $0.035 on Claude Fable 5 and $0.0105 on Kimi K3; repository review costs $0.65 and $0.195; the cache-heavy agent loop costs $0.9 and $0.27. Costs use the listed standard API rates.
Which has the larger context window, Claude Fable 5 or Kimi K3?
Kimi K3 has the larger documented context window: 1.05M, compared with 1M.
Related comparisons
Last updated September 4, 2026
Watch Claude Fable 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.