# GPT-5.5 vs Hy3 Preview > Side-by-side benchmark comparison for GPT-5.5 and Hy3 Preview across agentic, coding, multimodal, knowledge, reasoning, multilingual, and math tasks. - Canonical page: https://benchlm.ai/compare/gpt-5-5-vs-hy3-preview - Last updated: September 23, 2026 - Shared sourced benchmarks: 5 - HTML indexing: indexable - Ranking lane: BenchAlign v5.6 ## Quick Verdict Pick GPT-5.5 if you want the stronger benchmark profile. Hy3 Preview only makes more sense when its price, context window, or workload-specific category wins matter more than the overall score. ## Summary - GPT-5.5 leads overall 69.48 to 45.32. - The clearest category separation is in instruction following, where the averages are 91.9 for GPT-5.5 and 46.9 for Hy3 Preview. - The biggest single benchmark swing is Gert Labs in Agentic, with scores of 72.93% and 36.91%. - Hy3 Preview is the cheaper option on output tokens, which matters if you expect large responses or heavy interactive use. - GPT-5.5 also has the larger context window at 1M. ## Model Snapshot | Property | GPT-5.5 | Hy3 Preview | |----------|----------|----------| | Creator | OpenAI | Tencent | | Type | Proprietary | Open Weight | | Reasoning | Reasoning | Reasoning | | Context | 1M | 256K | | Overall Score | 69.48 | 45.32 | | Benchmarks Covered | 39 | 6 | | Pricing (input/output) | $5.00 / $30.00 | $0.00 / $0.00 | ## Category Breakdown ### Agentic - Winner: Not comparable - GPT-5.5 public-lane score: 60.2 (Supported · #16/105) - Hy3 Preview public-lane score: Coming soon | Benchmark | GPT-5.5 | Hy3 Preview | Winner | |-----------|-----------|-----------|--------| | Terminal-Bench 2.0 | 82% | 54.4% | GPT-5.5 | | CyberGym | 81.8% | Coming soon | Coming soon | | BrowseComp | 84.4% | Coming soon | Coming soon | | OSWorld-Verified | 78.7% | Coming soon | Coming soon | | MCP Atlas | 75.3% | Coming soon | Coming soon | | Toolathlon | 55.6% | Coming soon | Coming soon | | τ²-bench results | 98% | Coming soon | Coming soon | | Gert Labs | 72.93% | 36.91% | GPT-5.5 | | ResearchClawBench | 17.0% | Coming soon | Coming soon | | OSWorld 2.0 | 13.0% | Coming soon | Coming soon | | JobBench | 42.7% | Coming soon | Coming soon | | ExploitGym | 13.4% | Coming soon | Coming soon | | Terminal-Bench 2.1 (Vals) | 76.4% | Coming soon | Coming soon | | ApprenticeBench | 20% | Coming soon | Coming soon | ### Coding - Winner: Directional only - GPT-5.5 public-lane score: 64.5 (Supported · #10/135) - Hy3 Preview public-lane score: 33.3 (Estimated · #85/135) | Benchmark | GPT-5.5 | Hy3 Preview | Winner | |-----------|-----------|-----------|--------| | SWE-bench Pro | 58.6% | Coming soon | Coming soon | | Terminal-Bench 2.0 | 82.0% | 54.4% | GPT-5.5 | | Vibe Code Bench | 69.85% | Coming soon | Coming soon | | React Native Evals | 84.7% | Coming soon | Coming soon | | cursorBench31 | 59.2% | Coming soon | Coming soon | | cursorBench32 | 58.4% | Coming soon | Coming soon | | FrontierCode 1.1 Main | 43.0% | Coming soon | Coming soon | | LiveCodeBench (Vals) | 85.3% | Coming soon | Coming soon | | SWE-bench (Vals) | 82.6% | Coming soon | Coming soon | | SWE-bench Verified | Coming soon | 74.4% | Coming soon | ### Reasoning - Winner: Not comparable - GPT-5.5 public-lane score: 63.8 (#12/18) - Hy3 Preview public-lane score: Coming soon | Benchmark | GPT-5.5 | Hy3 Preview | Winner | |-----------|-----------|-----------|--------| | MRCR v2 64K-128K | 83.1% | Coming soon | Coming soon | | MRCR v2 128K-256K | 87.5% | Coming soon | Coming soon | | ARC-AGI-2 | 85% | Coming soon | Coming soon | | ARC-AGI-3 | 0.4% | Coming soon | Coming soon | ### Multimodal & Grounded - Winner: Not comparable - GPT-5.5 public-lane score: 71.4 (#20/50) - Hy3 Preview public-lane score: Coming soon | Benchmark | GPT-5.5 | Hy3 Preview | Winner | |-----------|-----------|-----------|--------| | MMMU-Pro | 81.2% | Coming soon | Coming soon | | MMMU-Pro w/ Python | 83.2% | Coming soon | Coming soon | | OfficeQA Pro | 54.1% | Coming soon | Coming soon | ### Knowledge - Winner: Directional only - GPT-5.5 public-lane score: 70.6 (Supported · #10/160) - Hy3 Preview public-lane score: 39.1 (Estimated · #93/160) | Benchmark | GPT-5.5 | Hy3 Preview | Winner | |-----------|-----------|-----------|--------| | GPQA | 93.6% | 87.2% | GPT-5.5 | | GPQA-D | 93.6% | 87.2% | GPT-5.5 | | HLE | 52.2% | Coming soon | Coming soon | | HLE w/o tools | 41.4% | Coming soon | Coming soon | | GPQA Diamond (Vals) | 93.2% | Coming soon | Coming soon | | MMLU-Pro (Vals) | 88.1% | Coming soon | Coming soon | ### Instruction Following - Winner: Directional only - GPT-5.5 public-lane score: 91.9 (#7/124) - Hy3 Preview public-lane score: 46.9 (#86/124) ### Mathematics - Winner: Not comparable - GPT-5.5 public-lane score: 69.4 (Unranked · 3 rankable rows) - Hy3 Preview public-lane score: Coming soon | Benchmark | GPT-5.5 | Hy3 Preview | Winner | |-----------|-----------|-----------|--------| | FrontierMath (legacy) | 51.7% | Coming soon | Coming soon | | FrontierMath v2 (Tiers 1-3) | 51.700% | Coming soon | Coming soon | | FrontierMath v2 (Tier 4) | 35.400% | Coming soon | Coming soon | ## FAQ ### Which is better overall, GPT-5.5 or Hy3 Preview? GPT-5.5 is ahead overall on BenchLM right now. ### Where is the biggest gap between GPT-5.5 and Hy3 Preview? The widest category gap is in instruction following, where the averages are 91.9 for GPT-5.5 and 46.9 for Hy3 Preview. ### How many benchmarks does GPT-5.5 cover on BenchLM? GPT-5.5 currently has 39 sourced benchmark scores on BenchLM. ### How many benchmarks does Hy3 Preview cover on BenchLM? Hy3 Preview currently has 6 sourced benchmark scores on BenchLM. ## Related Comparisons - [GPT-5.5 vs GPT-5.5 Pro](/compare/gpt-5-5-vs-gpt-5-5-pro) - [Hy3 Preview vs GPT-5.5 Pro](/compare/gpt-5-5-pro-vs-hy3-preview) - [GPT-5.5 vs Hy3](/compare/gpt-5-5-vs-hy3) - [Hy3 Preview vs Hy3](/compare/hy3-vs-hy3-preview) ## Explore More - [GPT-5.5 profile](/models/gpt-5-5) - [Hy3 Preview profile](/models/hy3-preview) - [Compare Pricing](/llm-pricing) - [Alternative Finder](/tools/alternative-finder) - [LLM Selector](/tools/llm-selector) - [Overall Rankings](/best/overall)