# GPT-6 Astra vs Hy4 preview

> Side-by-side benchmark comparison for GPT-6 Astra and Hy4 preview across agentic, coding, multimodal, knowledge, reasoning, multilingual, and math tasks.

- Canonical page: https://benchlm.ai/compare/gpt-6-astra-vs-hy4-preview
- Last updated: October 17, 2026

- Shared sourced benchmarks: 11
- HTML indexing: indexable
- Ranking lane: BenchAlign v5.8

## Quick Verdict

GPT-6 Astra leads by point estimate. Conditional score ranges do not establish rank confidence. Choose using the category evidence and your own workload tests.

## Summary

- GPT-6 Astra has the higher overall point estimate 83.88 to 61.27. Conditional score ranges do not establish rank confidence.
- The clearest category separation is in knowledge, where the averages are 86.2 for GPT-6 Astra and 64.1 for Hy4 preview.
- The biggest single benchmark swing is Agents' Last Exam in Agentic, with scores of 59.3% and 22.8%.
- Hy4 preview is the cheaper option on output tokens, which matters if you expect large responses or heavy interactive use.
- GPT-6 Astra also has the larger context window at 1.05M.

## Model Snapshot

| Property | GPT-6 Astra | Hy4 preview |
|----------|----------|----------|
| Creator | OpenAI | Tencent |
| Type | Proprietary | Open Weight |
| Reasoning | Reasoning | Reasoning |
| Context | 1.05M | 1M |
| Overall Score | 83.88 | 61.27 |
| Benchmarks Covered | 40 | 33 |
| Pricing (input/output) | $10.00 / $50.00 | $0.00 / $0.00 |

## Category Breakdown

### Agentic

- Winner: Directional only
- GPT-6 Astra public-lane score: 74.8 (Supported · #7/128)
- Hy4 preview public-lane score: 55.8 (Estimated · #40/128)

| Benchmark | GPT-6 Astra | Hy4 preview | Winner |
|-----------|-----------|-----------|--------|
| BrowseComp | 91.5% | Coming soon | Coming soon |
| OSWorld 2.0 | 72.6% | Coming soon | Coming soon |
| Terminal-Bench 4.0 | 57.90% | Coming soon | Coming soon |
| Terminal-Bench-Science 0.1 | 64.6% | Coming soon | Coming soon |
| ExploitGym | 42.4% | Coming soon | Coming soon |
| Agents' Last Exam | 59.3% | 22.8% | GPT-6 Astra |
| AutomationBench | 41.4% | 32.1% | GPT-6 Astra |
| HLE w/ tools | 57.2% | 55.4% | GPT-6 Astra |
| Terminal-Bench 2.1 (Vals) | 87.3% | 55.1% | GPT-6 Astra |
| ApprenticeBench | 68% | Coming soon | Coming soon |
| CWE-bench v1 | 68.0% | 53.0% | GPT-6 Astra |
| Hermes Index | 56.3% | 26.1% | GPT-6 Astra |
| Hermes Bench | 73.7% | 58.6% | GPT-6 Astra |
| MazeBench | 23% | Coming soon | Coming soon |
| Terminal-Bench 2.1 | Coming soon | 85.4% | Coming soon |
| CyberGym | Coming soon | 78.4% | Coming soon |
| WideResearch | Coming soon | 83.9% | Coming soon |
| DRACO | Coming soon | 77.2% | Coming soon |
| MCP Atlas | Coming soon | 83.7% | Coming soon |
| Toolathlon-Verified | Coming soon | 74.1% | Coming soon |
| APEX-Agents | Coming soon | 37.1% | Coming soon |
| skillsBench | Coming soon | 62.9% | Coming soon |
| JobBench | Coming soon | 61.7% | Coming soon |
| BankerToolBench | Coming soon | 78.6% | Coming soon |

### Coding

- Winner: GPT-6 Astra
- GPT-6 Astra public-lane score: 75.5 (Supported · #4/150)
- Hy4 preview public-lane score: 56.7 (Supported · #27/150)

| Benchmark | GPT-6 Astra | Hy4 preview | Winner |
|-----------|-----------|-----------|--------|
| DeepSWE | 74.1% | 64.3% | GPT-6 Astra |
| FrontierCode 1.1 Main | 53.3% | Coming soon | Coming soon |
| FrontierCode 1.1 Extended | 64.5% | Coming soon | Coming soon |
| FrontierSWE v2 | 65.5% | Coming soon | Coming soon |
| PostTrainBench v1.1 | 44.3% | Coming soon | Coming soon |
| Bug Hunt Bench | 45.0 fixes | Coming soon | Coming soon |
| Terminal-Bench 2.1 | Coming soon | 85.4% | Coming soon |
| SWE-bench Pro | Coming soon | 65.7% | Coming soon |
| SWE Multilingual | Coming soon | 82.9% | Coming soon |
| NL2Repo | Coming soon | 58.9% | Coming soon |
| ProgramBench | Coming soon | 17.5% | Coming soon |
| PostTrain Bench | Coming soon | 35.6% | Coming soon |
| sweMarathon | Coming soon | 31.9% | Coming soon |

### Reasoning

- Winner: Not comparable
- GPT-6 Astra public-lane score: 93.3 (#1/28)
- Hy4 preview public-lane score: Coming soon

| Benchmark | GPT-6 Astra | Hy4 preview | Winner |
|-----------|-----------|-----------|--------|
| STEMBench | 2.81 | 2.59 | GPT-6 Astra |
| ARC-AGI-1 | 98.50% | Coming soon | Coming soon |
| ARC-AGI-2 | 95% | Coming soon | Coming soon |
| ARC-AGI-3 | 62.7% | Coming soon | Coming soon |
| GeneBench-Pro | 37.8% | Coming soon | Coming soon |
| MRCR v2 256K-512K | 100.0% | Coming soon | Coming soon |
| MRCR v2 512K-1M | 96.3% | Coming soon | Coming soon |
| GraphWalks BFS 256K–1M | 71.8% | Coming soon | Coming soon |
| CritPt | Coming soon | 16.9% | Coming soon |

### Multimodal & Grounded

- Winner: Not comparable
- GPT-6 Astra public-lane score: 87.3 (Unranked · 3 rankable rows)
- Hy4 preview public-lane score: 88.3 (Unranked · 1 rankable row)

| Benchmark | GPT-6 Astra | Hy4 preview | Winner |
|-----------|-----------|-----------|--------|
| ScreenSpot Pro | 92.7% | Coming soon | Coming soon |
| BenchCAD Vision2Code (tools) | 0.959 | Coming soon | Coming soon |
| OfficeQA Pro | Coming soon | 66.2% | Coming soon |

### Knowledge

- Winner: Directional only
- GPT-6 Astra public-lane score: 86.2 (Supported · #3/184)
- Hy4 preview public-lane score: 64.1 (Estimated · #40/184)

| Benchmark | GPT-6 Astra | Hy4 preview | Winner |
|-----------|-----------|-----------|--------|
| GPQA | 96% | 92.3% | GPT-6 Astra |
| GPQA-D | 96.0% | 92.3% | GPT-6 Astra |
| HealthBench (raw) | 56.9% | Coming soon | Coming soon |
| HealthBench (length-adjusted) | 58.3% | Coming soon | Coming soon |
| HealthBench Professional | 64.7% | Coming soon | Coming soon |
| HealthBench Professional (raw) | 68.2% | Coming soon | Coming soon |
| HealthBench Hard | 36.6% | Coming soon | Coming soon |
| HLE | Coming soon | 55.4% | Coming soon |
| HLE w/o tools | Coming soon | 43.4% | Coming soon |

### Instruction Following

- Winner: Not comparable
- GPT-6 Astra public-lane score: Coming soon
- Hy4 preview public-lane score: Coming soon

| Benchmark | GPT-6 Astra | Hy4 preview | Winner |
|-----------|-----------|-----------|--------|
| Gray Swan IPI (15 attempts) | 8.5% | Coming soon | Coming soon |

### Mathematics

- Winner: Not comparable
- GPT-6 Astra public-lane score: 97.3 (Unranked · 2 rankable rows)
- Hy4 preview public-lane score: Coming soon

| Benchmark | GPT-6 Astra | Hy4 preview | Winner |
|-----------|-----------|-----------|--------|
| FrontierMath v2 (Tier 4) | 97.600% | Coming soon | Coming soon |
| FrontierMath v2 (Tiers 1-3) | 93.684% | Coming soon | Coming soon |
| Apex | Coming soon | 74.2% | Coming soon |

## FAQ

### Which is better overall, GPT-6 Astra or Hy4 preview?

GPT-6 Astra has the higher overall point estimate. Conditional score ranges do not establish rank confidence.

### Where is the biggest gap between GPT-6 Astra and Hy4 preview?

The widest category gap is in knowledge, where the averages are 86.2 for GPT-6 Astra and 64.1 for Hy4 preview.

### How many benchmarks does GPT-6 Astra cover on BenchLM?

GPT-6 Astra currently has 40 sourced benchmark scores on BenchLM.

### How many benchmarks does Hy4 preview cover on BenchLM?

Hy4 preview currently has 33 sourced benchmark scores on BenchLM.

## Related Comparisons

- [GPT-6 Astra vs GPT-6.1 Sol](/compare/gpt-6-1-sol-vs-gpt-6-astra)
- [Hy4 preview vs GPT-6.1 Sol](/compare/gpt-6-1-sol-vs-hy4-preview)
- [GPT-6 Astra vs GPT-6 Sol](/compare/gpt-6-astra-vs-gpt-6-sol)
- [Hy4 preview vs GPT-6 Sol](/compare/gpt-6-sol-vs-hy4-preview)

## Explore More

- [GPT-6 Astra profile](/models/gpt-6-astra)
- [Hy4 preview profile](/models/hy4-preview)
- [Compare Pricing](/llm-pricing)
- [Alternative Finder](/tools/alternative-finder)
- [LLM Selector](/tools/llm-selector)
- [Overall Rankings](/best/overall)