Long documents
Prompts that approach the documented context limit
Hy4
Hy4 has the larger documented context window.
Confidence: documented
Hy4 vs Kimi K2.5
Updated November 4, 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. 2 results are shared. Category rows resting on Estimated evidence or different benchmark sets are marked directional and do not name a winner.
- Shared results
- 2
- Hy4 only
- 13
- Kimi K2.5 only
- 66
- Like-for-like categories
- 0 / 8
Supported: Kimi K2.5How 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.
Chat turn cost
1K fresh input + 500 output tokens
Hy4
Hy4 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
Hy4
Hy4 has the lower estimated token cost for this stated workload. Kimi K2.5 has no published cached-input rate, so cached tokens use its listed input rate.
Confidence: rate-fallback
Show secondary and unsupported calls
Repository review cost
50K fresh input + 3K output tokens
Kimi K2.5
Kimi K2.5 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
Hy4 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
Hy4 is not ranked on the public lane for agentic, so no winner is named for agentic.
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.
—Hy439.3Kimi K2.5
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.
13 Hy4 only
2 Shared
66 Kimi K2.5 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 29.0Hy4:59.1%Kimi K2.5:30.1%MMMU-ProMultimodal
Normalized gap 0.1Hy4:78.6%Kimi K2.5:78.5%
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.
Agentic
Not comparable- Hy4
- Not ranked
- Kimi K2.5
- 42.3
- Estimated · #70/135
- Basis
- BenchAlign v5.8 lane · 7 vs 16 public rows
- Reading
- Not comparable
Coding
Not comparable- Hy4
- Not ranked
- Kimi K2.5
- 39.3
- Estimated · #69/156
- Basis
- BenchAlign v5.8 lane · 5 vs 10 public rows
- Reading
- Not comparable
Reasoning
Not comparable- Hy4
- Not ranked
- Kimi K2.5
- 71.3
- Unranked · 3 rankable rows
- Basis
- Provisional lane · 0 vs 1 weighted rows
- Reading
- Not comparable
Multimodal
Not comparable- Hy4
- Not ranked
- Kimi K2.5
- 70.4
- #36/58
- Basis
- Provisional lane · 1 vs 2 weighted rows
- Reading
- Not comparable
Knowledge
Not comparable- Hy4
- Not ranked
- Kimi K2.5
- 50.5
- Estimated · #79/179
- Basis
- BenchAlign v5.8 lane · 1 vs 8 public rows
- Reading
- Not comparable
Multilingual
Not comparable- Hy4
- Not ranked
- Kimi K2.5
- 87.3
- #8/17
- Basis
- Provisional lane · 0 vs 1 weighted rows
- Reading
- Not comparable
Instruction following
Not comparable- Hy4
- Not ranked
- Kimi K2.5
- 85.8
- #51/157
- Basis
- Provisional lane · 0 vs 0 weighted rows
- Reading
- Not comparable
Math
Not comparable- Hy4
- Not ranked
- Kimi K2.5
- 83.3
- Unranked · 7 rankable rows
- Basis
- Provisional lane · 0 vs 2 weighted rows
- Reading
- Not comparable
| Category | Hy4 | Kimi K2.5 | Basis | Reading |
|---|---|---|---|---|
| Agentic | Not ranked | 42.3Estimated · #70/135 | Not comparableBenchAlign v5.8 lane · 7 vs 16 public rows | Not comparable |
| Coding | Not ranked | 39.3Estimated · #69/156 | Not comparableBenchAlign v5.8 lane · 5 vs 10 public rows | Not comparable |
| Reasoning | Not ranked | 71.3Unranked · 3 rankable rows | Not comparableProvisional lane · 0 vs 1 weighted rows | Not comparable |
| Multimodal | Not ranked | 70.4#36/58 | Not comparableProvisional lane · 1 vs 2 weighted rows | Not comparable |
| Knowledge | Not ranked | 50.5Estimated · #79/179 | Not comparableBenchAlign v5.8 lane · 1 vs 8 public rows | Not comparable |
| Multilingual | Not ranked | 87.3#8/17 | Not comparableProvisional lane · 0 vs 1 weighted rows | Not comparable |
| Instruction following | Not ranked | 85.8#51/157 | Not comparableProvisional lane · 0 vs 0 weighted rows | Not comparable |
| Math | Not ranked | 83.3Unranked · 7 rankable rows | Not comparableProvisional lane · 0 vs 2 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.
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
- Hy4
- $0.00208
- Fits in one request
- Kimi K2.5
- $0.0021
- Fits in one request
Hy4 has the lower modeled cost
Costs use the listed standard API rates.
Repository review
50K fresh input + 3K output tokens
- Hy4
- $0.0492
- Fits in one request
- Kimi K2.5
- $0.039
- Fits in one request
Kimi K2.5 has the lower modeled cost
Costs use the listed standard API rates.
Cache-heavy agent loop
200K cached + 20K fresh input + 10K output tokens
- Hy4
- $0.05009
- Fits in one request
- Kimi K2.5
- $0.162
- Fits in one request
- Cached input priced at the published list-input rate
Hy4 has the lower modeled cost
Kimi K2.5 has no published cached-input rate, so cached tokens use its listed input rate.
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.
SpecificationHy4Kimi K2.5
Context window
Maximum documented context; output-token limits may be lower.
Kimi K2.5
256K
API model ID
Hy4
tencent/hy4
Tencent Cloud TokenHub model pricingKimi K2.5
Not sourced
Cached-input rate
A missing cached-input rate falls back to the listed input rate only in the stated workload estimate.
Hy4
$0.042 per 1M cached input tokens
Tencent Cloud TokenHub model pricingKimi K2.5
Not published
Documented inputs
Hy4
Not sourced
Kimi K2.5
Not sourced
Documented outputs
Hy4
Not sourced
Kimi K2.5
Not sourced
Provider availability
Hy4
Not sourced
Kimi K2.5
Not sourced
Reasoning profile
Hy4
Reasoning
Kimi K2.5
Non-Reasoning
Weight access
Hy4
Pending
Kimi K2.5
Open Weight
License
Hy4
Pending
Kimi K2.5
Open Weight
Release date
Hy4
2026-11-04
Kimi K2.5
2026-01-27
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
- 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: $0.0492 vs $0.039. Cache-heavy agent loop: $0.05009 vs $0.162.
- Context tradeoff
- Hy4 has the larger documented window (1M).
Run the same representative tasks against both endpoints before changing production traffic.
Questions
Which is better, Hy4 or Kimi K2.5?
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, Hy4 or Kimi K2.5?
Hy4 is not ranked on the public lane for coding, so no winner is named for coding.
Which is better for agentic tasks, Hy4 or Kimi K2.5?
Hy4 is not ranked on the public lane for agentic tasks, so no winner is named for agentic tasks.
Which costs less, Hy4 or Kimi K2.5?
For the stated presets, chat costs $0.00209 on Hy4 and $0.0021 on Kimi K2.5; repository review costs $0.0492 and $0.039; the cache-heavy agent loop costs $0.05009 and $0.162. Kimi K2.5 has no published cached-input rate, so cached tokens use its listed input rate.
Which has the larger context window, Hy4 or Kimi K2.5?
Hy4 has the larger documented context window: 1M, compared with 256K.
Self-host vs API cost
Estimates at 50,000 req/day · 1000 tokens/req average.
Hy4
API / mo$2,501
Self-host / moNot listed
Break-even—
Proprietary model — self-hosting not applicable.
Kimi K2.5
API / mo$2,700
Self-host / mo$27,962
Break-even706M/day
Benchmark evidence
The full public result ledger is available for audit without forcing a wide desktop table onto a phone.
Browse raw public benchmark evidence81 rows
Agentic
Terminal-Bench 4.0
Hy444.40%SourceKimi K2.5—Not directly comparable
Terminal-Bench-Science 0.1
Hy419.8%SourceKimi K2.5—Not directly comparable
HLE w/ tools
Hy459.1%SourceKimi K2.5—Not directly comparable
Agents' Last Exam
Hy441.0%SourceKimi K2.5—Not directly comparable
APEX-Agents
Hy475.2%SourceKimi K2.5—Not directly comparable
AutomationBench
Hy456.2%SourceKimi K2.5—Not directly comparable
OSWorld 2.1
Hy478.9%SourceKimi K2.5—Not directly comparable
Terminal-Bench 2.0
Hy4—Kimi K2.550.8%SourceNot directly comparable
BrowseComp
Hy4—Kimi K2.560.6%SourceNot directly comparable
OSWorld-Verified
Hy4—Kimi K2.563.3%SourceNot directly comparable
Claw-Eval
Hy4—Kimi K2.552.3%SourceNot directly comparable
QwenClawBench
Hy4—Kimi K2.554.3%SourceNot directly comparable
τ³-bench results
Hy4—Kimi K2.565.7%SourceNot directly comparable
DeepSearchQA
Hy4—Kimi K2.577.1%SourceNot directly comparable
DeepPlanning
Hy4—Kimi K2.514.4%SourceNot directly comparable
Toolathlon
Hy4—Kimi K2.527.8%SourceNot directly comparable
MCP Atlas
Hy4—Kimi K2.529.5%SourceNot directly comparable
MCP-Tasks
Hy4—Kimi K2.559.1%SourceNot directly comparable
WideResearch
Hy4—Kimi K2.572.7%SourceNot directly comparable
BFCL v4
Hy4—Kimi K2.568.3%SourceNot directly comparable
Gert Labs
Hy4—Kimi K2.545.88%SourceNot directly comparable
ResearchClawBench
Hy4—Kimi K2.514.0%SourceNot directly comparable
JobBench
Hy4—Kimi K2.58.7%SourceNot directly comparable
Coding
DeepSWE
Hy474.9%SourceKimi K2.5—Not directly comparable
sweMarathon
Hy443.8%SourceKimi K2.5—Not directly comparable
ProgramBench
Hy430.0%SourceKimi K2.5—Not directly comparable
FrontierCode 2.0
Hy431.0%SourceKimi K2.5—Not directly comparable
PostTrainBench v1.2
Hy435.2%SourceKimi K2.5—Not directly comparable
SWE-bench Verified
Hy4—Kimi K2.576.8%SourceNot directly comparable
SWE-bench Verified*
Hy4—Kimi K2.570.8%SourceNot directly comparable
LiveCodeBench v6
Hy4—Kimi K2.585.0%SourceNot directly comparable
SWE-bench Pro
Hy4—Kimi K2.550.7%SourceNot directly comparable
SWE Multilingual
Hy4—Kimi K2.573%SourceNot directly comparable
SWE-Rebench
Hy4—Kimi K2.558.5%SourceNot directly comparable
React Native Evals
Hy4—Kimi K2.577.2%SourceNot directly comparable
NL2Repo
Hy4—Kimi K2.532%SourceNot directly comparable
SciCode
Hy4—Kimi K2.548.7%SourceNot directly comparable
Terminal-Bench 2.0
Hy4—Kimi K2.550.8%SourceNot directly comparable
Reasoning
LongBench v2
Hy4—Kimi K2.561%SourceNot directly comparable
Multimodal
MMMU-Pro
Hy478.6%SourceKimi K2.578.5%SourceHy4 leads this result
ZeroBench w/ Python
Hy455.4%SourceKimi K2.5—Not directly comparable
MMMU
Hy4—Kimi K2.584.3%SourceNot directly comparable
RealWorldQA
Hy4—Kimi K2.581.0%SourceNot directly comparable
OmniDocBench 1.5
Hy4—Kimi K2.588.8%SourceNot directly comparable
Video-MME
Hy4—Kimi K2.587.4%SourceNot directly comparable
Video-MME (with subtitle)
Hy4—Kimi K2.587.4%SourceNot directly comparable
Video-MME (w/o subtitle)
Hy4—Kimi K2.583.2%SourceNot directly comparable
MathVision
Hy4—Kimi K2.584.2%SourceNot directly comparable
We-Math
Hy4—Kimi K2.584.7%SourceNot directly comparable
DynaMath
Hy4—Kimi K2.584.4%SourceNot directly comparable
SimpleVQA
Hy4—Kimi K2.571.2%SourceNot directly comparable
MMLongBench-Doc
Hy4—Kimi K2.558.5%SourceNot directly comparable
CC-OCR
Hy4—Kimi K2.579.7%SourceNot directly comparable
AI2D_TEST
Hy4—Kimi K2.590.8%SourceNot directly comparable
CountBench
Hy4—Kimi K2.594.1%SourceNot directly comparable
RefCOCO (avg)
Hy4—Kimi K2.587.8%SourceNot directly comparable
MMVU
Hy4—Kimi K2.580.4%SourceNot directly comparable
VideoMMMU
Hy4—Kimi K2.586.6%SourceNot directly comparable
MLVU (M-Avg)
Hy4—Kimi K2.585.0%SourceNot directly comparable
CharXiv
Hy4—Kimi K2.577.5%SourceNot directly comparable
Knowledge
HLE
Hy459.1%SourceKimi K2.530.1%SourceHy4 leads this result
GPQA
Hy4—Kimi K2.587.6%SourceNot directly comparable
GPQA-D
Hy4—Kimi K2.587.6%SourceNot directly comparable
SuperGPQA
Hy4—Kimi K2.569.2%SourceNot directly comparable
MMLU-Pro
Hy4—Kimi K2.587.1%SourceNot directly comparable
MMLU-Pro (Arcee)
Hy4—Kimi K2.587.1%SourceNot directly comparable
MMLU-Redux
Hy4—Kimi K2.594.5%SourceNot directly comparable
C-Eval
Hy4—Kimi K2.594%SourceNot directly comparable
Multilingual
Instruction following
IFEval
Hy4—Kimi K2.593.9%SourceNot directly comparable
Math
AIME 2025
Hy4—Kimi K2.596.1%SourceNot directly comparable
AIME26
Hy4—Kimi K2.595.8%SourceNot directly comparable
AIME25 (Arcee)
Hy4—Kimi K2.596.3%SourceNot directly comparable
HMMT Feb 2025
Hy4—Kimi K2.595.4%SourceNot directly comparable
HMMT Nov 2025
Hy4—Kimi K2.591.1%SourceNot directly comparable
HMMT Feb 2026
Hy4—Kimi K2.587.1%SourceNot directly comparable
MMAnswerBench
Hy4—Kimi K2.581.8%SourceNot directly comparable
81 public results · 2 shared
Watch Hy4 vs Kimi K2.5
One weekly email when material rank, price, or benchmark evidence changes make this matchup worth revisiting.
Read a sample issueJoin 6,100+ subscribers.
Last updated November 4, 2026