Long documents
Prompts that approach the documented context limit
GPT-5.4
GPT-5.4 has the larger documented context window.
Confidence: documented
GPT-5.4 vs Hy3 Preview
Updated September 23, 2026. Rank says GPT-5.4 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
GPT-5.4 has the higher public score, 69.23 versus 45.32, and the 90% score intervals do not overlap. 4 results are shared. Category rows resting on Estimated evidence or different benchmark sets are marked directional and do not name a winner.
- Shared results
- 4
- GPT-5.4 only
- 34
- Hy3 Preview only
- 2
- Like-for-like categories
- 0 / 8
Supported: GPT-5.4 · Estimated: Hy3 PreviewHow 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.
Show secondary and unsupported calls
Coding work
Code generation, repair, and software-engineering tasks
Not enough matched evidence
Hy3 Preview is scored on Estimated evidence for coding, so the reading is directional rather than like-for-like.
Confidence: limitedAgentic work
Tool use, computer use, and multi-step task completion
Not enough matched evidence
Hy3 Preview is not ranked on the public lane for agentic, so no winner is named for agentic.
Confidence: limitedChat 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-ratesCache-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-ratesRepository 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
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.
51.0GPT-5.433.3Hy3 Preview
Directional only · BenchAlign v5.6
GPT-5.4 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.
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.
34 GPT-5.4 only
4 Shared
2 Hy3 Preview only
3 categories rest 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.
Terminal-Bench 2.0Agentic
Normalized gap 20.7GPT-5.4:75.1%Hy3 Preview:54.4%GPQAKnowledge
Normalized gap 5.6GPT-5.4:92.8%Hy3 Preview: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 v5.6 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- GPT-5.4
- 51.0
- Supported · #38/135
- Hy3 Preview
- 33.3
- Estimated · #85/135
- Basis
- BenchAlign v5.6 lane · 4 vs 2 public rows
- Reading
- Directional only
Knowledge
Directional only- GPT-5.4
- 66.5
- Supported · #17/160
- Hy3 Preview
- 39.1
- Estimated · #93/160
- Basis
- BenchAlign v5.6 lane · 7 vs 2 public rows
- Reading
- Directional only
Instruction following
Directional only- GPT-5.4
- 89.2
- #19/124
- Hy3 Preview
- 46.9
- #86/124
- Basis
- Provisional lane · 0 vs 0 weighted rows
- Reading
- Directional only
Agentic
Not comparable- GPT-5.4
- 49.5
- Supported · #36/105
- Hy3 Preview
- Not ranked
- Basis
- BenchAlign v5.6 lane · 14 vs 2 public rows
- Reading
- Not comparable
Reasoning
Not comparable- GPT-5.4
- 57.2
- #15/18
- Hy3 Preview
- Not ranked
- Basis
- Provisional lane · 2 vs 0 weighted rows
- Reading
- Not comparable
Multimodal
Not comparable- GPT-5.4
- 69.3
- #21/50
- Hy3 Preview
- Not ranked
- Basis
- Provisional lane · 3 vs 0 weighted rows
- Reading
- Not comparable
Multilingual
Not comparable- GPT-5.4
- Not ranked
- Hy3 Preview
- Not ranked
- Basis
- Provisional lane · 0 vs 0 weighted rows
- Reading
- Not comparable
Math
Not comparable- GPT-5.4
- 64.4
- Unranked · 2 rankable rows
- Hy3 Preview
- Not ranked
- Basis
- Provisional lane · 2 vs 0 weighted rows
- Reading
- Not comparable
| Category | GPT-5.4 | Hy3 Preview | Basis | Reading |
|---|---|---|---|---|
| Coding | 51.0Supported · #38/135 | 33.3Estimated · #85/135 | Directional onlyBenchAlign v5.6 lane · 4 vs 2 public rows | Directional only |
| Knowledge | 66.5Supported · #17/160 | 39.1Estimated · #93/160 | Directional onlyBenchAlign v5.6 lane · 7 vs 2 public rows | Directional only |
| Instruction following | 89.2#19/124 | 46.9#86/124 | Directional onlyProvisional lane · 0 vs 0 weighted rows | Directional only |
| Agentic | 49.5Supported · #36/105 | Not ranked | Not comparableBenchAlign v5.6 lane · 14 vs 2 public rows | Not comparable |
| Reasoning | 57.2#15/18 | Not ranked | Not comparableProvisional lane · 2 vs 0 weighted rows | Not comparable |
| Multimodal | 69.3#21/50 | Not ranked | Not comparableProvisional lane · 3 vs 0 weighted rows | Not comparable |
| Multilingual | Not ranked | Not ranked | Not comparableProvisional lane · 0 vs 0 weighted rows | Not comparable |
| Math | 64.4Unranked · 2 rankable rows | Not ranked | Not comparableProvisional lane · 2 vs 0 weighted rows | Not comparable |
Ranks count the models scored in each category’s lane, so the agentic, coding, and knowledge denominators (BenchAlign v5.6) 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
- GPT-5.4
- $0.01
- Fits in one request
- Hy3 Preview
- Self-hosted; infrastructure cost varies
- Fits in one request
Hy3 Preview has no comparable published API token rate.
Repository review
50K fresh input + 3K output tokens
- GPT-5.4
- $0.17
- Fits in one request
- Hy3 Preview
- Self-hosted; infrastructure cost varies
- Fits in one request
Hy3 Preview has no comparable published API token rate.
Cache-heavy agent loop
200K cached + 20K fresh input + 10K output tokens
- GPT-5.4
- $0.25
- Fits in one request
- Hy3 Preview
- Self-hosted; infrastructure cost varies
- Fits in one request
- Cached-input rate unavailable
Hy3 Preview has no comparable published API token 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.
SpecificationGPT-5.4Hy3 Preview
Context window
Maximum documented context; output-token limits may be lower.
GPT-5.4
1.05M
OpenAI pricingHy3 Preview
256K
API model ID
GPT-5.4
gpt-5.4
OpenAI pricingHy3 Preview
Not sourced
Cached-input rate
A missing cached-input rate falls back to the listed input rate only in the stated workload estimate.
GPT-5.4
$0.25 per 1M cached input tokens
OpenAI pricingHy3 Preview
No comparable hosted API rate
Documented inputs
GPT-5.4
Not sourced
Hy3 Preview
Not sourced
Documented outputs
GPT-5.4
Not sourced
Hy3 Preview
Not sourced
Provider availability
GPT-5.4
Not sourced
Hy3 Preview
Not sourced
Reasoning profile
GPT-5.4
Reasoning
Hy3 Preview
Reasoning
Weight access
GPT-5.4
Proprietary
Hy3 Preview
Open Weight
License
GPT-5.4
Proprietary
Hy3 Preview
Open Weight
Release date
GPT-5.4
2026-03-05
Hy3 Preview
2026-04-23
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
- GPT-5.4 has the higher public score, 69.23 versus 45.32, and the 90% score intervals do not overlap.
- Workload cost
- A complete comparable API-rate estimate is not available for both models.
- Context tradeoff
- GPT-5.4 has the larger documented window (1.05M).
Run the same representative tasks against both endpoints before changing production traffic.
Questions
Which is better, GPT-5.4 or Hy3 Preview?
GPT-5.4 has the higher public score, 69.23 versus 45.32, 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, GPT-5.4 or Hy3 Preview?
GPT-5.4 scores higher for coding on the public lane, 51 to 33.3. Hy3 Preview 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, GPT-5.4 or Hy3 Preview?
Hy3 Preview is not ranked on the public lane for agentic tasks, so no winner is named for agentic tasks.
Which costs less, GPT-5.4 or Hy3 Preview?
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, GPT-5.4 or Hy3 Preview?
GPT-5.4 has the larger documented context window: 1.05M, compared with 256K.
Benchmark evidence
The full public result ledger is available for audit without forcing a wide desktop table onto a phone.
Browse raw public benchmark evidence40 rows
Agentic
Terminal-Bench 2.0
GPT-5.475.1%SourceHy3 Preview54.4%SourceGPT-5.4 leads this result
CyberGym
GPT-5.479.0%SourceHy3 Preview—Not directly comparable
BrowseComp
GPT-5.482.7%SourceHy3 Preview—Not directly comparable
OSWorld-Verified
GPT-5.475%SourceHy3 Preview—Not directly comparable
MCP Atlas
GPT-5.470.6%SourceHy3 Preview—Not directly comparable
Toolathlon
GPT-5.454.6%SourceHy3 Preview—Not directly comparable
τ²-bench results
GPT-5.498.9%SourceHy3 Preview—Not directly comparable
Claw-Eval
GPT-5.460.3%SourceHy3 Preview—Not directly comparable
DeepSearchQA
GPT-5.473.6%SourceHy3 Preview—Not directly comparable
Gert Labs
Shared sourceGPT-5.464.89%Hy3 Preview36.91%GPT-5.4 leads this result
ResearchClawBench
GPT-5.415.3%SourceHy3 Preview—Not directly comparable
JobBench
GPT-5.438.9%SourceHy3 Preview—Not directly comparable
ExploitGym
GPT-5.46.0%SourceHy3 Preview—Not directly comparable
ApprenticeBench
GPT-5.411%SourceHy3 Preview—Not directly comparable
Coding
LiveCodeBench Pro
GPT-5.487.5%SourceHy3 Preview—Not directly comparable
SWE-bench Pro
GPT-5.457.7%SourceHy3 Preview—Not directly comparable
React Native Evals
GPT-5.485.3%SourceHy3 Preview—Not directly comparable
Vibe Code Bench
GPT-5.467.42%SourceHy3 Preview—Not directly comparable
SWE-bench Verified
GPT-5.4—Hy3 Preview74.4%SourceNot directly comparable
Terminal-Bench 2.0
GPT-5.4—Hy3 Preview54.4%SourceNot directly comparable
Reasoning
Multimodal
MMMU-Pro
GPT-5.481.2%SourceHy3 Preview—Not directly comparable
OfficeQA Pro
GPT-5.453.2%SourceHy3 Preview—Not directly comparable
MMMU-Pro w/ Python
GPT-5.482.1%SourceHy3 Preview—Not directly comparable
CharXiv
GPT-5.482.8%SourceHy3 Preview—Not directly comparable
ERQA
GPT-5.465.4%SourceHy3 Preview—Not directly comparable
SimpleVQA
GPT-5.461.1%SourceHy3 Preview—Not directly comparable
ScreenSpot Pro
GPT-5.485.4%SourceHy3 Preview—Not directly comparable
ZeroBench
GPT-5.441.0%SourceHy3 Preview—Not directly comparable
MedXpertQA (MM)
GPT-5.477.1%SourceHy3 Preview—Not directly comparable
Knowledge
GPQA
GPT-5.492.8%SourceHy3 Preview87.2%SourceGPT-5.4 leads this result
HLE
GPT-5.452.1%SourceHy3 Preview—Not directly comparable
HLE w/o tools
GPT-5.439.8%SourceHy3 Preview—Not directly comparable
GPQA-D
GPT-5.492.8%SourceHy3 Preview87.2%SourceGPT-5.4 leads this result
HealthBench Hard
GPT-5.440.1%SourceHy3 Preview—Not directly comparable
MedXpertQA (Text)
GPT-5.459.6%SourceHy3 Preview—Not directly comparable
HealthBench Professional
GPT-5.448.1%SourceHy3 Preview—Not directly comparable
Math
40 public results · 4 shared
Watch GPT-5.4 vs Hy3 Preview
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 23, 2026