GPT-6 Sol vs Hy4 preview
Updated October 21, 2026. 21 benchmarks have a result for both. Coding is ranked on the Vibe Coding Index; every other board is shown as published.
Decision reading
Pick GPT-6 Sol for coding: it scores 82.8 on the Vibe Coding Index against 61.6 for Hy4 preview.
Pick Hy4 preview when output tokens cost 4.0x less, or when what matters is Finance Agent (Vals AI) or Tax Agent Bench (Vals AI), where it leads.
- Shared results
- 21
- GPT-6 Sol only
- 79
- Hy4 preview only
- 3
- Like-for-like categories
- 5 of 9
Shared results can support a head-to-head reading. Results only one model has describe coverage, not superiority.
Which one for your work
Each card is decided by one number both models carry. Where only one has it, the card says so and makes no call.
Coding
Vibe Coding Index, four public coding benchmarks
GPT-6 Sol
GPT-6 Sol scores 82.8 against 61.6.
2 and 1 benchmarks behind the scores
Intelligence
AA Intelligence Index
No call
Only GPT-6 Sol has a AA Intelligence Index result, 47.6.
Directional only
Web design
Design Arena: websites
No call
Only GPT-6 Sol has a Design Arena: websites result, 1292.
Directional only
General
Vals Index
No call
Only Hy4 preview has a Vals Index result, 53.2%.
Directional only
Long documents
Listed context window
GPT-6 Sol
GPT-6 Sol takes 1.05M tokens against 1.05M.
Listed by OpenRouter
Everyday cost
1K input + 500 output tokens
Hy4 preview
GPT-6 Sol $0.0070, Hy4 preview $0.0021 per turn.
List prices
At a glance
- Coding index
- GPT-6 Sol 82.8 from 2 benchmarks, Hy4 preview 61.6 from 1 benchmark.
- Widest category
- Agentic work, 16.5 points apart on average over 7 shared benchmarks.
- Widest single gap
- Terminal-Bench 4 (Vals AI): GPT-6 Sol 44.4% against Hy4 preview 8.1%, 36.4 points apart on a 0 to 100 scale.
- Output price
- GPT-6 Sol $10.00 against Hy4 preview $2.501 per million output tokens, 4.0x apart.
- Context window
- GPT-6 Sol takes 1.05M tokens, Hy4 preview 1.05M.
Benchmark by benchmark
First the score: boards each model leads, per category, where both have a result. Then every board, a category to a fold; bars are each result on a 0 to 100 scale, so compare the two in a row, not one row with another.
| Category | GPT-6 Sol | Bars | Hy4 preview | Ties |
|---|---|---|---|---|
| Coding | 10 | 0 | ||
| Agentic work | 4 | 3 | ||
| Math | 1 | 0 | ||
| Knowledge | 1 | 0 | ||
| General | 0 | 2 | ||
| All shared boards | 16 | 5 |
CodingLike-for-likeFour of these boards feed the Vibe Coding Index100
| Benchmark | GPT-6 Sol | Hy4 preview | Leader |
|---|---|---|---|
| LiveBench coding | 93.1% | Not measured | Only GPT-6 Sol measured |
| LMArena WebDev | 1683 | 1631 | GPT-6 Sol |
| Terminal-Bench 4 (Vals AI) | 44.4% | 8.1% | GPT-6 Sol |
| Vibe Code Bench (Vals AI) | 87.8% | 77.5% | GPT-6 Sol |
| Code migration (Vals AI) | 57.2% | 47.4% | GPT-6 Sol |
| ProgramBench (Vals AI) | 2.0% | 0.0% | GPT-6 Sol |
| IOI (Vals AI) | 82.6% | 59.3% | GPT-6 Sol |
| Terminal-Bench 2.1 (Vals AI) | 83.2% | 55.1% | GPT-6 Sol |
| CyberBench (Vals AI) | 78.0% | 65.4% | GPT-6 Sol |
| SciCode | 57.6% | Not measured | Only GPT-6 Sol measured |
| FrontierCode | 49.3% | Not measured | Only GPT-6 Sol measured |
| ALE-Bench | 2462.0 | Not measured | Only GPT-6 Sol measured |
| LMArena Text: coding | 1448 | Not measured | Only GPT-6 Sol measured |
| LMArena Text: software and IT | 1433 | Not measured | Only GPT-6 Sol measured |
| LMArena WebDev: React | 1685 | 1653 | GPT-6 Sol |
| LMArena WebDev: HTML | 1685 | 1620 | GPT-6 Sol |
| LMArena WebDev: from an image | 1660 | Not measured | Only GPT-6 Sol measured |
| Neither model has a result on the other 21 boards here | |||
IntelligenceDirectional onlyPublic boards, shown as published, kept out of the index2 boards, none shared
| Benchmark | GPT-6 Sol | Hy4 preview | Leader |
|---|---|---|---|
| AA Intelligence Index | 47.6 | Not measured | Only GPT-6 Sol measured |
| Epoch Capabilities Index | 163.7 | Not measured | Only GPT-6 Sol measured |
Agentic workLike-for-likePublic boards, shown as published, kept out of the index43
| Benchmark | GPT-6 Sol | Hy4 preview | Leader |
|---|---|---|---|
| Design Arena: agents building web apps | 1232 | Not measured | Only GPT-6 Sol measured |
| Design Arena: agents building full-stack apps | 1238 | Not measured | Only GPT-6 Sol measured |
| Design Arena: agents building mobile apps | 1216 | Not measured | Only GPT-6 Sol measured |
| LiveBench research | 69.3% | Not measured | Only GPT-6 Sol measured |
| LiveBench browser | 92.0% | Not measured | Only GPT-6 Sol measured |
| LiveBench long agentic | 66.9% | Not measured | Only GPT-6 Sol measured |
| SREBench (Vals AI) | 30.5% | 2.3% | GPT-6 Sol |
| Finance Agent (Vals AI) | 49.0% | 55.1% | Hy4 preview |
| Tax Agent Bench (Vals AI) | 15.0% | 32.6% | Hy4 preview |
| Legal Research Bench (Vals AI) | 28.9% | 45.2% | Hy4 preview |
| Harvey's Legal Agent Benchmark (Vals AI) | Not measured | 8.8% | Only Hy4 preview measured |
| Excel Modeling Benchmark (Vals AI) | 71.5% | 58.1% | GPT-6 Sol |
| BioMysteryBench (Vals AI) | 74.8% | 69.3% | GPT-6 Sol |
| Terminal-Bench-Science (Vals AI) | 30.0% | 1.4% | GPT-6 Sol |
| Vals Science Index | 45.0% | Not measured | Only GPT-6 Sol measured |
| APEX-Agents | 54.3% | Not measured | Only GPT-6 Sol measured |
| SpireClimb | 7.5% | Not measured | Only GPT-6 Sol measured |
| Vending-Bench 2 | $14,428 | Not measured | Only GPT-6 Sol measured |
| Neither model has a result on the other 21 boards here | |||
ReasoningDirectional onlyPublic boards, shown as published, kept out of the index14 boards, none shared
| Benchmark | GPT-6 Sol | Hy4 preview | Leader |
|---|---|---|---|
| LiveBench reasoning | 88.7% | Not measured | Only GPT-6 Sol measured |
| MysteryMechanism (Vals AI) | 30.2% | Not measured | Only GPT-6 Sol measured |
| Humanity's Last Exam (Artificial Analysis run) | 47.9% | Not measured | Only GPT-6 Sol measured |
| GPQA Diamond (Epoch AI run) | 94.3% | Not measured | Only GPT-6 Sol measured |
| ARC-AGI | 95.5% | Not measured | Only GPT-6 Sol measured |
| ARC-AGI-2 | 89.6% | Not measured | Only GPT-6 Sol measured |
| Chess puzzles | 56.0% | Not measured | Only GPT-6 Sol measured |
| Mystery game puzzles | 56.0% | Not measured | Only GPT-6 Sol measured |
| EBR-bench | 53.3% | Not measured | Only GPT-6 Sol measured |
| LMArena Text: hard prompts | 1422 | Not measured | Only GPT-6 Sol measured |
| LMArena Text: expert | 1437 | Not measured | Only GPT-6 Sol measured |
| LMArena Text: hard prompts in English | 1420 | Not measured | Only GPT-6 Sol measured |
| LMArena Vision: diagrams | 1270 | Not measured | Only GPT-6 Sol measured |
| LMArena Vision: homework | 1286 | Not measured | Only GPT-6 Sol measured |
| Neither model has a result on the other 7 boards here | |||
MathLike-for-likePublic boards, shown as published, kept out of the index10
| Benchmark | GPT-6 Sol | Hy4 preview | Leader |
|---|---|---|---|
| OTIS Mock AIME | 100.0% | Not measured | Only GPT-6 Sol measured |
| FrontierMath, tiers 1 to 3 | 89.8% | Not measured | Only GPT-6 Sol measured |
| FrontierMath, tier 4 | 90.0% | Not measured | Only GPT-6 Sol measured |
| ProofBench | 83.0% | 75.0% | GPT-6 Sol |
| LMArena Text: math | 1407 | Not measured | Only GPT-6 Sol measured |
| LMArena Text: mathematical work | 1424 | Not measured | Only GPT-6 Sol measured |
| Neither model has a result on the other 7 boards here | |||
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index10
| Benchmark | GPT-6 Sol | Hy4 preview | Leader |
|---|---|---|---|
| LegalBench (Vals AI) | Not measured | 83.8% | Only Hy4 preview measured |
| MedCode (Vals AI) | 47.1% | 43.3% | GPT-6 Sol |
| SimpleQA Verified | 60.7% | Not measured | Only GPT-6 Sol measured |
| LMArena Text: science | 1389 | Not measured | Only GPT-6 Sol measured |
| LMArena Text: medicine and healthcare | 1366 | Not measured | Only GPT-6 Sol measured |
| LMArena Text: legal and government | 1396 | Not measured | Only GPT-6 Sol measured |
| LMArena Text: business and finance | 1380 | Not measured | Only GPT-6 Sol measured |
| Neither model has a result on the other 10 boards here | |||
Web designDirectional onlyPublic boards, shown as published, kept out of the index9 boards, none shared
| Benchmark | GPT-6 Sol | Hy4 preview | Leader |
|---|---|---|---|
| Design Arena: websites | 1292 | Not measured | Only GPT-6 Sol measured |
| Design Arena: UI components | 1292 | Not measured | Only GPT-6 Sol measured |
| Design Arena: game dev | 1313 | Not measured | Only GPT-6 Sol measured |
| Design Arena: data viz | 1269 | Not measured | Only GPT-6 Sol measured |
| Design Arena: 3D | 1322 | Not measured | Only GPT-6 Sol measured |
| Design Arena: all code categories | 1301 | Not measured | Only GPT-6 Sol measured |
| Design Arena: SVG | 1318 | Not measured | Only GPT-6 Sol measured |
| Design Arena: ASCII art | 1226 | Not measured | Only GPT-6 Sol measured |
| LiveBench design | 82.2% | Not measured | Only GPT-6 Sol measured |
GeneralLike-for-likePublic boards, shown as published, kept out of the index02
| Benchmark | GPT-6 Sol | Hy4 preview | Leader |
|---|---|---|---|
| Vals Index | Not measured | 53.2% | Only Hy4 preview measured |
| LiveBench data analysis | 81.2% | Not measured | Only GPT-6 Sol measured |
| Public Benefits Bench (Vals AI) | 56.6% | 68.6% | Hy4 preview |
| SAGE (Vals AI) | 44.8% | Not measured | Only GPT-6 Sol measured |
| MedScribe (Vals AI) | 82.0% | 83.6% | Hy4 preview |
| Long context reasoning (AA-LCR) | 83.7% | Not measured | Only GPT-6 Sol measured |
| DTBench | 97.3% | Not measured | Only GPT-6 Sol measured |
| LMCA | 59.1% | Not measured | Only GPT-6 Sol measured |
| Furniture assembly | 58.3% | Not measured | Only GPT-6 Sol measured |
| LMArena Text | 1398 | Not measured | Only GPT-6 Sol measured |
| LMArena Text: instruction following | 1413 | Not measured | Only GPT-6 Sol measured |
| LMArena Text: creative writing | 1378 | Not measured | Only GPT-6 Sol measured |
| LMArena Vision | 1240 | Not measured | Only GPT-6 Sol measured |
| LMArena Text: English | 1396 | Not measured | Only GPT-6 Sol measured |
| LMArena Text: non-English | 1388 | Not measured | Only GPT-6 Sol measured |
| LMArena Text: Chinese | 1412 | Not measured | Only GPT-6 Sol measured |
| LMArena Text: German | 1379 | Not measured | Only GPT-6 Sol measured |
| LMArena Text: Japanese | 1377 | Not measured | Only GPT-6 Sol measured |
| LMArena Text: Korean | 1336 | Not measured | Only GPT-6 Sol measured |
| LMArena Text: French | 1399 | Not measured | Only GPT-6 Sol measured |
| LMArena Text: Spanish | 1387 | Not measured | Only GPT-6 Sol measured |
| LMArena Text: Russian | 1402 | Not measured | Only GPT-6 Sol measured |
| LMArena Text: Polish | 1372 | Not measured | Only GPT-6 Sol measured |
| LMArena Text: multi-turn | 1414 | Not measured | Only GPT-6 Sol measured |
| LMArena Text: longer queries | 1416 | Not measured | Only GPT-6 Sol measured |
| LMArena Text: writing and language | 1389 | Not measured | Only GPT-6 Sol measured |
| LMArena Text: entertainment and media | 1373 | Not measured | Only GPT-6 Sol measured |
| LMArena Vision: English | 1229 | Not measured | Only GPT-6 Sol measured |
| LMArena Vision: reading text | 1258 | Not measured | Only GPT-6 Sol measured |
| LMArena Vision: creative writing | 1219 | Not measured | Only GPT-6 Sol measured |
| Neither model has a result on the other 16 boards here | |||
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
| Benchmark | GPT-6 Sol | Hy4 preview | Leader |
|---|---|---|---|
| Neither model has a result on 5 boards here | |||
What each workload costs
Three fixed token mixes at each model's listed price, so per-token rates become a bill.
Chat turn
1K input + 500 output tokens
Hy4 preview costs less
Repository review
50K input + 3K output tokens
Hy4 preview costs less
Agent loop
200K input + 10K output tokens
Hy4 preview costs less
Listed OpenRouter rates, without cache or batch discounts. A workload that does not fit a model's context window would need splitting.
Specification differences
The leading value in a row is marked; a row with a missing value names no leader. Our two rows are marked in red.
| Specification | OpenAI GPT-6 Sol | Tencent Hy4 preview |
|---|---|---|
| Maker | OpenAI | Tencent |
| Weights | No public weights listed | Open weights |
| Released | September 22, 2026 | August 28, 2026 |
| Context window | 1.05M | 1.05M |
| Input $/Mtok | $2.00 | $0.834 |
| Output $/Mtok | $10.00 | $2.501 |
| Coding index | 82.8 | 61.6 |
| Independent measurements behind it | 2 | 1 |
| Cost per solved task | Not computable | Not computable |
| Source | OpenRouter listing | OpenRouter listing |
Questions
Is GPT-6 Sol better than Hy4 preview for coding?
Pick GPT-6 Sol for coding: it scores 82.8 on the Vibe Coding Index against 61.6 for Hy4 preview. Pick Hy4 preview when output tokens cost 4.0x less, or when what matters is Finance Agent (Vals AI) or Tax Agent Bench (Vals AI), where it leads.
Which is cheaper, GPT-6 Sol or Hy4 preview?
GPT-6 Sol costs $2.00 per million input tokens and $10.00 per million output tokens; Hy4 preview costs $0.834 and $2.501. Cost per solved task is known for neither of them.
Which has the bigger context window, GPT-6 Sol or Hy4 preview?
GPT-6 Sol, with 1.05M tokens against 1.05M.
How many benchmarks is the GPT-6 Sol vs Hy4 preview comparison based on?
21 benchmarks have a result for both models as of October 21, 2026. GPT-6 Sol's coding index rests on 2 independent benchmarks and Hy4 preview's on 1. Where only one model has a result, the page says so instead of naming a winner.