Agentic
Directional only- Hy4 preview
- 58.6
- Estimated · #30/154
- Ternary Bonsai 2 27B
- 49.9
- Estimated · #59/154
- Basis
- BenchAlign lane · 13 vs 3 public rows
- Reading
- Directional only
Hy4 preview vs Ternary Bonsai 2 27B
Decision reading
Hy4 preview has the higher public score estimate, 60.89 versus 50.78, but the 90% score intervals overlap. Treat that as a lead, not a settled winner.
4 results are shared. Category rows resting on Estimated evidence or different benchmark sets are marked directional and do not name a winner.
Updated September 18, 2026. Rank says Hy4 preview 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.
Long documents
Prompts that approach the documented context limit
Hy4 preview
Hy4 preview has the larger documented context window.
Confidence: documented
Show secondary and unsupported calls
Coding work
Code generation, repair, and software-engineering tasks
Not enough matched evidence
Hy4 preview and Ternary Bonsai 2 27B are scored on Estimated evidence for coding, so the reading is directional rather than like-for-like.
Confidence: limited
Agentic work
Tool use, computer use, and multi-step task completion
Not enough matched evidence
Hy4 preview and Ternary Bonsai 2 27B are scored on Estimated evidence for agentic, so the reading is directional rather than like-for-like.
Confidence: limited
Chat turn cost
1K fresh input + 500 output tokens
Not enough matched evidence
A complete comparable API-rate estimate is not available for both models.
Confidence: listed-rates
Cache-heavy agent loop cost
200K cached + 20K fresh input + 10K output tokens
Not enough matched evidence
A complete comparable API-rate estimate is not available for both models.
Confidence: listed-rates
Repository review cost
50K fresh input + 3K output tokens
Not enough matched evidence
A complete comparable API-rate estimate is not available for both models.
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.
24 Hy4 preview only
4 Shared
17 Ternary Bonsai 2 27B only
- Shared results
- 4
- Hy4 preview only
- 24
- Ternary Bonsai 2 27B only
- 17
- Like-for-like categories
- 0 / 8
3 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- Hy4 preview
- 59.4
- Estimated · #24/154
- Ternary Bonsai 2 27B
- 49.9
- Estimated · #64/154
- Basis
- BenchAlign lane · 8 vs 4 public rows
- Reading
- Directional only
Knowledge
Directional only- Hy4 preview
- 56.8
- Estimated · #45/184
- Ternary Bonsai 2 27B
- 50.6
- Estimated · #78/184
- Basis
- BenchAlign lane · 4 vs 3 public rows
- Reading
- Directional only
Reasoning
Not comparable- Hy4 preview
- Not ranked
- Ternary Bonsai 2 27B
- 73.9
- Unranked · 1 rankable row
- Basis
- Provisional lane · 0 vs 0 weighted rows
- Reading
- Not comparable
Math
Not comparable- Hy4 preview
- Not ranked
- Ternary Bonsai 2 27B
- 76.8
- Unranked · 4 rankable rows
- Basis
- Provisional lane · 0 vs 1 weighted rows
- Reading
- Not comparable
Multilingual
Not comparable- Hy4 preview
- Not ranked
- Ternary Bonsai 2 27B
- Not ranked
- Basis
- Provisional lane · 0 vs 0 weighted rows
- Reading
- Not comparable
Multimodal
Not comparable- Hy4 preview
- 84.9
- Unranked · 1 rankable row
- Ternary Bonsai 2 27B
- Not ranked
- Basis
- Provisional lane · 1 vs 0 weighted rows
- Reading
- Not comparable
Instruction following
Not comparable- Hy4 preview
- Not ranked
- Ternary Bonsai 2 27B
- 71.0
- #64/124
- Basis
- Provisional lane · 0 vs 1 weighted rows
- Reading
- Not comparable
| Category | Hy4 preview | Ternary Bonsai 2 27B | Basis | Reading |
|---|---|---|---|---|
| Agentic | 58.6Estimated · #30/154 | 49.9Estimated · #59/154 | Directional onlyBenchAlign lane · 13 vs 3 public rows | Directional only |
| Coding | 59.4Estimated · #24/154 | 49.9Estimated · #64/154 | Directional onlyBenchAlign lane · 8 vs 4 public rows | Directional only |
| Knowledge | 56.8Estimated · #45/184 | 50.6Estimated · #78/184 | Directional onlyBenchAlign lane · 4 vs 3 public rows | Directional only |
| Reasoning | Not ranked | 73.9Unranked · 1 rankable row | Not comparableProvisional lane · 0 vs 0 weighted rows | Not comparable |
| Math | Not ranked | 76.8Unranked · 4 rankable rows | Not comparableProvisional lane · 0 vs 1 weighted rows | Not comparable |
| Multilingual | Not ranked | Not ranked | Not comparableProvisional lane · 0 vs 0 weighted rows | Not comparable |
| Multimodal | 84.9Unranked · 1 rankable row | Not ranked | Not comparableProvisional lane · 1 vs 0 weighted rows | Not comparable |
| Instruction following | Not ranked | 71.0#64/124 | Not comparableProvisional lane · 0 vs 1 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.
GPQA
Knowledge
Hy4 preview: 92.3%Ternary Bonsai 2 27B: 85.8%Normalized gap 6.5Hy4 preview source Ternary Bonsai 2 27B 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
- Hy4 preview
- Self-hosted; infrastructure cost varies
- Fits in one request
- Ternary Bonsai 2 27B
- Self-hosted; infrastructure cost varies
- Fits in one request
Hy4 preview has no comparable published API token rate. Ternary Bonsai 2 27B has no comparable published API token rate.
Repository review
50K fresh input + 3K output tokens
- Hy4 preview
- Self-hosted; infrastructure cost varies
- Fits in one request
- Ternary Bonsai 2 27B
- Self-hosted; infrastructure cost varies
- Fits in one request
Hy4 preview has no comparable published API token rate. Ternary Bonsai 2 27B has no comparable published API token rate.
Cache-heavy agent loop
200K cached + 20K fresh input + 10K output tokens
- Hy4 preview
- Self-hosted; infrastructure cost varies
- Fits in one request
- Cached-input rate unavailable
- Ternary Bonsai 2 27B
- Self-hosted; infrastructure cost varies
- Fits in one request
- Cached-input rate unavailable
Hy4 preview has no comparable published API token rate. Ternary Bonsai 2 27B has no comparable published API token rate.
Specification differences
Sourced differences are shown directly. Missing facts stay explicit instead of being inferred from a model name or family.
SpecificationHy4 previewTernary Bonsai 2 27B
Context window
Maximum documented context; output-token limits may be lower.
Hy4 preview
Ternary Bonsai 2 27B
API model ID
Hy4 preview
tencent/Hy4-preview
Tencent Hy4 preview model cardTernary Bonsai 2 27B
prism-ml/Ternary-Bonsai-2-27B-gguf
PrismML Bonsai 2 collectionCached-input rate
A missing cached-input rate falls back to the listed input rate only in the stated workload estimate.
Hy4 preview
No comparable hosted API rate
Tencent Hy4 preview model cardTernary Bonsai 2 27B
No comparable hosted API rate
PrismML Bonsai 2 collectionDocumented inputs
Hy4 preview
Not sourced
Ternary Bonsai 2 27B
Not sourced
Documented outputs
Hy4 preview
Not sourced
Ternary Bonsai 2 27B
Not sourced
Provider availability
Hy4 preview
Not sourced
Ternary Bonsai 2 27B
Not sourced
Reasoning profile
Hy4 preview
Reasoning
Ternary Bonsai 2 27B
Reasoning
Weight access
Hy4 preview
Open Weight
Ternary Bonsai 2 27B
Open Weight
License
Hy4 preview
Open Weight
Ternary Bonsai 2 27B
Open Weight
Release date
Hy4 preview
2026-08-28
Ternary Bonsai 2 27B
2026-09-17
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
- Hy4 preview has the higher public score estimate, 60.89 versus 50.78, but the 90% score intervals overlap.
- Workload cost
- A complete comparable API-rate estimate is not available for both models.
- Context tradeoff
- Hy4 preview has the larger documented window (1M).
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 evidence45 rows
Agentic
Terminal-Bench 2.1
Hy4 preview85.4%SourceTernary Bonsai 2 27B52.8%SourceHy4 preview leads this result
CyberGym
Hy4 preview78.4%SourceTernary Bonsai 2 27B—Not directly comparable
WideResearch
Hy4 preview83.9%SourceTernary Bonsai 2 27B—Not directly comparable
DRACO
Hy4 preview77.2%SourceTernary Bonsai 2 27B—Not directly comparable
MCP Atlas
Hy4 preview83.7%SourceTernary Bonsai 2 27B—Not directly comparable
Toolathlon-Verified
Hy4 preview74.1%SourceTernary Bonsai 2 27B—Not directly comparable
APEX-Agents
Hy4 preview37.1%SourceTernary Bonsai 2 27B—Not directly comparable
skillsBench
Hy4 preview62.9%SourceTernary Bonsai 2 27B—Not directly comparable
JobBench
Hy4 preview61.7%SourceTernary Bonsai 2 27B—Not directly comparable
Agents' Last Exam
Hy4 preview22.8%SourceTernary Bonsai 2 27B—Not directly comparable
AutomationBench
Hy4 preview32.1%SourceTernary Bonsai 2 27B—Not directly comparable
BankerToolBench
Hy4 preview78.6%SourceTernary Bonsai 2 27B—Not directly comparable
HLE w/ tools
Hy4 preview55.4%SourceTernary Bonsai 2 27B—Not directly comparable
τ²-bench results
Hy4 preview—Ternary Bonsai 2 27B80.2%SourceNot directly comparable
BFCL v3
Hy4 preview—Ternary Bonsai 2 27B74.9%SourceNot directly comparable
Coding
Terminal-Bench 2.1
Hy4 preview85.4%SourceTernary Bonsai 2 27B52.8%SourceHy4 preview leads this result
SWE-bench Pro
Hy4 preview65.7%SourceTernary Bonsai 2 27B—Not directly comparable
SWE Multilingual
Hy4 preview82.9%SourceTernary Bonsai 2 27B—Not directly comparable
DeepSWE
Hy4 preview64.3%SourceTernary Bonsai 2 27B—Not directly comparable
NL2Repo
Hy4 preview58.9%SourceTernary Bonsai 2 27B—Not directly comparable
ProgramBench
Hy4 preview17.5%SourceTernary Bonsai 2 27B—Not directly comparable
PostTrain Bench
Hy4 preview35.6%SourceTernary Bonsai 2 27B—Not directly comparable
sweMarathon
Hy4 preview31.9%SourceTernary Bonsai 2 27B—Not directly comparable
LiveCodeBench v6
Hy4 preview—Ternary Bonsai 2 27B90.1%SourceNot directly comparable
BigCodeBench
Hy4 preview—Ternary Bonsai 2 27B58.1%SourceNot directly comparable
SWE-bench Verified
Hy4 preview—Ternary Bonsai 2 27B60.8%SourceNot directly comparable
Reasoning
CritPt
Hy4 preview16.9%SourceTernary Bonsai 2 27B—Not directly comparable
Knowledge
GPQA
Hy4 preview92.3%SourceTernary Bonsai 2 27B85.8%SourceHy4 preview leads this result
GPQA-D
Hy4 preview92.3%SourceTernary Bonsai 2 27B85.8%SourceHy4 preview leads this result
HLE
Hy4 preview55.4%SourceTernary Bonsai 2 27B—Not directly comparable
HLE w/o tools
Hy4 preview43.4%SourceTernary Bonsai 2 27B—Not directly comparable
MMLU-Redux
Hy4 preview—Ternary Bonsai 2 27B89.1%SourceNot directly comparable
Math
Apex
Hy4 preview74.2%SourceTernary Bonsai 2 27B—Not directly comparable
GSM8K
Hy4 preview—Ternary Bonsai 2 27B96.7%SourceNot directly comparable
MATH-500
Hy4 preview—Ternary Bonsai 2 27B98.8%SourceNot directly comparable
AIME 2025
Hy4 preview—Ternary Bonsai 2 27B95%SourceNot directly comparable
AIME26
Hy4 preview—Ternary Bonsai 2 27B95.8%SourceNot directly comparable
Multimodal
OfficeQA Pro
Hy4 preview66.2%SourceTernary Bonsai 2 27B—Not directly comparable
CharXiv (overall)
Hy4 preview—Ternary Bonsai 2 27B80.0%SourceNot directly comparable
A-OKVQA
Hy4 preview—Ternary Bonsai 2 27B86.8%SourceNot directly comparable
OmniDocBench 1.6
Hy4 preview—Ternary Bonsai 2 27B89.1%SourceNot directly comparable
RealWorldQA
Hy4 preview—Ternary Bonsai 2 27B80.1%SourceNot directly comparable
OCRBench V2
Hy4 preview—Ternary Bonsai 2 27B56.9%SourceNot directly comparable
Instruction following
Questions
Which is better, Hy4 preview or Ternary Bonsai 2 27B?
Hy4 preview has the higher public score estimate, 60.89 versus 50.78, but the 90% score intervals overlap. The higher estimate is not a decisive winner because the uncertainty ranges overlap.
Which is better for coding, Hy4 preview or Ternary Bonsai 2 27B?
Hy4 preview scores higher for coding on the public lane, 59.4 to 49.9. Hy4 preview and Ternary Bonsai 2 27B are 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, Hy4 preview or Ternary Bonsai 2 27B?
Hy4 preview scores higher for agentic tasks on the public lane, 58.6 to 49.9. Hy4 preview and Ternary Bonsai 2 27B are scored on Estimated evidence for agentic tasks, so the reading is directional rather than like-for-like. BenchLM does not name a winner for agentic tasks from a directional reading; read the shared benchmark rows directly and test the models on the same task set.
Which costs less, Hy4 preview or Ternary Bonsai 2 27B?
Both models do not have comparable published API token rates, so this page does not name a universal price winner.
Which has the larger context window, Hy4 preview or Ternary Bonsai 2 27B?
Hy4 preview has the larger documented context window: 1M, compared with 262K.
Related comparisons
Last updated September 18, 2026
Watch Hy4 preview vs Ternary Bonsai 2 27B
One weekly email when material rank, price, or benchmark evidence changes make this matchup worth revisiting.
Read a sample issueJoin 2,000+ readers.