Compare / head-to-head
GPT-6 Lunavs
Hy4 preview
Hy4 preview leads 2 of 4 shared benchmarks. GPT-6 Luna is 5.6x cheaper per token. GPT-6 Luna has the larger context window (1.1M tokens).
Benchmarks from cited public sources; pricing from official pages; status from official provider feeds.
Shared benchmarks
1 – 2
Hy4 preview leads · 1 tied
Cheaper per token
GPT-6 Luna
5.6x cheaper, input + output
Larger context
GPT-6 Luna
1.1M tokens
Providers
2 providers
OpenAI · Tencent Hunyuan
GPT-6 Luna
OpenAI · released 2026-09-22
textvision
Hy4 preview
Tencent Hunyuan · released 2026-08-28
text
Quality
Benchmark matrix
4 shared · 18 only GPT-6 Luna · 14 only Hy4 preview| Benchmark | GPT-6 Luna | Hy4 preview | Δ | Edge |
|---|---|---|---|---|
| Reported by both · 4 | ||||
| GPQA Diamond | 90.5% ↗epoch run | 92.3% ↗ | -1.8 pt | Hy4 preview |
| LMArena Agent (LMArena) | 0.0135 ↗ | 0.0395 ↗ | 0 | Tie |
| LMArena WebDev (LMArena) | 1578.7 ↗ | 1633.2 ↗ | -54.5 | Hy4 preview |
| Terminal-Bench 4.0 (Vals AI) | 13.6% ↗ | 8.1% ↗ | +5.5 pt | GPT-6 Luna |
| Only GPT-6 Luna reports · 18 | ||||
| LMArena Elo | 1442.9 ↗ | not reported | — | — |
| APEX-Agents (Mercor) | 44.3% ↗ | not reported | — | — |
| Chess Puzzles (Epoch AI run) | 31% ↗epoch run | not reported | — | — |
| DeepSWE v1.1 | 66.6% ↗ | not reported | — | — |
| FrontierMath Tier 4 v2 (Epoch AI run) | 56.1% ↗epoch run | not reported | — | — |
| FrontierMath Tiers 1-3 v2 (Epoch AI run) | 78.9% ↗epoch run | not reported | — | — |
| Furniture Assembly (Epoch AI run) | 44.2% ↗epoch run | not reported | — | — |
| LiveBench Agentic Coding (LiveBench) | 51.2% ↗ | not reported | — | — |
| LiveBench Coding (LiveBench) | 79% ↗ | not reported | — | — |
| LiveBench Data Analysis (LiveBench) | 73.4% ↗ | not reported | — | — |
| LiveBench Instruction Following (LiveBench) | 55.9% ↗ | not reported | — | — |
| LiveBench Language (LiveBench) | 73.8% ↗ | not reported | — | — |
| LiveBench Mathematics (LiveBench) | 89.1% ↗ | not reported | — | — |
| LiveBench Reasoning (LiveBench) | 81.8% ↗ | not reported | — | — |
| Mystery Game Puzzles (Epoch AI run) | 7% ↗epoch run | not reported | — | — |
| OTIS Mock AIME 2024-2025 (Epoch AI run) | 98.9% ↗epoch run | not reported | — | — |
| SAGE (Vals AI) | 48.1% ↗ | not reported | — | — |
| SimpleQA Verified | 41.4% ↗epoch run | not reported | — | — |
| Only Hy4 preview reports · 14 | ||||
| Agents' Last Exam (ALE-CLI) | not reported | 22.8% ↗ | — | — |
| CyberGym | not reported | 78.4% ↗ | — | — |
| DeepSWE | not reported | 64.3% ↗ | — | — |
| HLE (no tools, text-only) | not reported | 43.4% ↗ | — | — |
| HLE (with tools, text-only) | not reported | 55.4% ↗ | — | — |
| MathArena Apex 2025 | not reported | 74.2% ↗ | — | — |
| MCP-Atlas (public) | not reported | 83.7% ↗ | — | — |
| NL2Repo-Bench | not reported | 58.9% ↗ | — | — |
| OfficeQA Pro | not reported | 66.2% ↗ | — | — |
| SWE-bench Multilingual | not reported | 82.9% ↗ | — | — |
| SWE-bench Pro | not reported | 65.7% ↗ | — | — |
| Terminal-Bench 2.1 | not reported | 85.4% ↗ | — | — |
| Toolathlon-Verified | not reported | 74.1% ↗ | — | — |
| WideSearch | not reported | 83.9% ↗ | — | — |
Scores tagged "epoch" or "matharena" are independent runs, used only where the lab has not published its own; ⚠ marks rows MathArena flags as released after the competition. Higher is better on every row. Δ is GPT-6 Luna minus Hy4 preview in the benchmark's own unit. "Not reported" means the lab has not published that figure; it is not a zero. ↗ opens the source.
Specs & pricing
Side by side
Official pricing pages and model cards| Spec | GPT-6 Luna | Hy4 preview | Edge |
|---|---|---|---|
| Context window | 1.1M tokens | 1M tokens | GPT-6 Luna |
| Max output | 128K tokens | 64K tokens | GPT-6 Luna |
| Input price / 1M | $0.1 ↗ | $0.834 ↗ | GPT-6 Luna |
| Output price / 1M | $0.5 ↗ | $2.5 ↗ | GPT-6 Luna |
| Cached input / 1M | $0.01 ↗ | $0.042 ↗ | GPT-6 Luna |
| Input + output / 1M Lower is cheaper. List prices; batch, tool and regional fees excluded. | $0.6 | $3.34 | GPT-6 Luna |
| Modalities | text · vision | text | GPT-6 Luna |
| Released | 2026-09-22 | 2026-08-28 | — |
| Cited benchmark scores | 22 | 18 | — |
Reliability
Provider status
Live from /statusMore matchups
GPT-6 Luna vs …
Models sharing the most benchmarksMore matchups
Hy4 preview vs …
Models sharing the most benchmarksBuilt by Respan
Which one wins on your data?
Public benchmarks are a starting point. Run GPT-6 Luna and Hy4 preview on your own prompts with Respan evals, or route to either through one gateway key with automatic failover.