Enter opens the highlighted result. Esc closes.

GPT-5.6 Sol vs Hy4 flash

GPT-5.6 Sol leads Hy4 flash on coding, 70.6 to 44.5 on the Vibe Coding Index, and Hy4 flash costs 142.3x less per output token ($0.1405 vs $20.00 per million). Updated November 22, 2026; 10 benchmarks have a result for both.

GPT-5.6 SolOpenAI
Coding
70.6
Overall
78.9
Output $/M
$20.00
10
Hy4 flashTencent
Coding
44.5
Overall
No data
Output $/M
$0.1405
0
vs boards won of 10 shared

GPT-5.6 Sol or Hy4 flash: which to pick

Pick GPT-5.6 Sol for coding: it scores 70.6 on the Vibe Coding Index against 44.5 for Hy4 flash.

Pick Hy4 flash for output tokens 142.3x cheaper ($0.1405 vs $20.00 per million).

Shared results
10
GPT-5.6 Sol only
138
Hy4 flash only
2
Like-for-like categories
3 of 9

Only benchmarks that tested both models count toward a pick; a result only one model has is shown but not compared.

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.

At a glance

Coding index
GPT-5.6 Sol 70.6 from 5 benchmarks, Hy4 flash 44.5 from 2 benchmarks.
Widest category
Coding, 21.8 points apart on average over 3 shared benchmarks.
Widest single gap
Terminal-Bench 4 (Vals AI): GPT-5.6 Sol 37.9% against Hy4 flash 15.2%, 30.3 points apart on a 0 to 100 scale.
Output price
GPT-5.6 Sol $20.00 against Hy4 flash $0.1405 per million output tokens, 142.3x apart.
Context window
GPT-5.6 Sol takes 1.05M tokens, Hy4 flash 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.

Boards won, where both have a result
CategoryGPT-5.6 SolBarsHy4 flashTies
Coding30
Agentic work20
Web design50
All shared boards100
CodingLike-for-likeSix of these boards feed the Vibe Coding Index30
Coding benchmarks for GPT-5.6 Sol and Hy4 flash
BenchmarkGPT-5.6 SolHy4 flashLeader
LMArena WebDev1615Not measuredOnly GPT-5.6 Sol measured
LiveCodeBench (Vals AI)82.6%Not measuredOnly GPT-5.6 Sol measured
SWE-bench Verified (Vals AI)96.2%Not measuredOnly GPT-5.6 Sol measured
Terminal-Bench 4 (Vals AI)37.9%15.2%GPT-5.6 Sol
Vibe Code Bench (Vals AI)80.5%74.8%GPT-5.6 Sol
Code migration (Vals AI)52.9%Not measuredOnly GPT-5.6 Sol measured
ProgramBench (Vals AI)1.5%Not measuredOnly GPT-5.6 Sol measured
IOI (Vals AI)91.2%Not measuredOnly GPT-5.6 Sol measured
Terminal-Bench 2.1 (Vals AI)85.8%Not measuredOnly GPT-5.6 Sol measured
Vibe Code Bench 1-100 (Vals AI)20.0%Not measuredOnly GPT-5.6 Sol measured
SkillsBench (Vals AI)54.1%Not measuredOnly GPT-5.6 Sol measured
CyberBench (Vals AI)76.3%Not measuredOnly GPT-5.6 Sol measured
SciCode57.8%Not measuredOnly GPT-5.6 Sol measured
Terminal-Bench Hard (Artificial Analysis run)65.9%Not measuredOnly GPT-5.6 Sol measured
Terminal-Bench 4.0 (Artificial Analysis run)39.9%Not measuredOnly GPT-5.6 Sol measured
FrontierCode47.5%Not measuredOnly GPT-5.6 Sol measured
FrontierSWE32.2%Not measuredOnly GPT-5.6 Sol measured
GSO-Bench76.5%Not measuredOnly GPT-5.6 Sol measured
WeirdML88.8%Not measuredOnly GPT-5.6 Sol measured
MirrorCode20.0%Not measuredOnly GPT-5.6 Sol measured
ALE-Bench2176.9Not measuredOnly GPT-5.6 Sol measured
LMArena Text: coding1495Not measuredOnly GPT-5.6 Sol measured
LMArena Text: software and IT1487Not measuredOnly GPT-5.6 Sol measured
LMArena WebDev: React1618Not measuredOnly GPT-5.6 Sol measured
LMArena WebDev: HTML1613Not measuredOnly GPT-5.6 Sol measured
LMArena WebDev: from an image1606Not measuredOnly GPT-5.6 Sol measured
SciCode10.8%Not measuredOnly GPT-5.6 Sol measured
SWE-Bench Pro V295.5%Not measuredOnly GPT-5.6 Sol measured
SWE Atlas: Codebase QnA46.0%Not measuredOnly GPT-5.6 Sol measured
SWE Atlas: Test Writing45.9%Not measuredOnly GPT-5.6 Sol measured
GameBench 280.150.5GPT-5.6 Sol
Terminal-Bench 3.034.6%Not measuredOnly GPT-5.6 Sol measured
Terminal-Bench 4.037.3%Not measuredOnly GPT-5.6 Sol measured
CursorBench 4.040.9%Not measuredOnly GPT-5.6 Sol measured
ProgramBench1.0%Not measuredOnly GPT-5.6 Sol measured
Neither model has a result on the other 17 boards here
IntelligenceDirectional onlyPublic boards, shown as published, kept out of the index1 board, none shared
Intelligence benchmarks for GPT-5.6 Sol and Hy4 flash
BenchmarkGPT-5.6 SolHy4 flashLeader
Epoch Capabilities Index163.0Not measuredOnly GPT-5.6 Sol measured
Neither model has a result on the other 1 board here
Agentic workLike-for-likePublic boards, shown as published, kept out of the index20
Agentic work benchmarks for GPT-5.6 Sol and Hy4 flash
BenchmarkGPT-5.6 SolHy4 flashLeader
Design Arena: agents building web apps1210Not measuredOnly GPT-5.6 Sol measured
Design Arena: agents building full-stack apps1236Not measuredOnly GPT-5.6 Sol measured
Design Arena: agents building mobile apps1224Not measuredOnly GPT-5.6 Sol measured
Design Arena: agents building Android apps1215Not measuredOnly GPT-5.6 Sol measured
Design Arena: agents building Godot games1227Not measuredOnly GPT-5.6 Sol measured
Design Arena: agents building games1211Not measuredOnly GPT-5.6 Sol measured
Design Arena: HTML slides1179Not measuredOnly GPT-5.6 Sol measured
Design Arena: PowerPoint slides1205Not measuredOnly GPT-5.6 Sol measured
Design Arena: spreadsheets1144Not measuredOnly GPT-5.6 Sol measured
SREBench (Vals AI)30.5%Not measuredOnly GPT-5.6 Sol measured
Finance Agent (Vals AI)53.8%Not measuredOnly GPT-5.6 Sol measured
Tax Agent Bench (Vals AI)31.1%23.3%GPT-5.6 Sol
Legal Research Bench (Vals AI)48.1%38.5%GPT-5.6 Sol
Harvey's Legal Agent Benchmark (Vals AI)Not measured5.4%Only Hy4 flash measured
Excel Modeling Benchmark (Vals AI)72.3%Not measuredOnly GPT-5.6 Sol measured
BioMysteryBench (Vals AI)71.1%Not measuredOnly GPT-5.6 Sol measured
Terminal-Bench-Science (Vals AI)20.0%Not measuredOnly GPT-5.6 Sol measured
Vals Science Index41.5%Not measuredOnly GPT-5.6 Sol measured
Time Horizon Index (Vals AI)23.8%Not measuredOnly GPT-5.6 Sol measured
CUA-bench (Vals AI)8.3%Not measuredOnly GPT-5.6 Sol measured
τ²-Bench Telecom85.1%Not measuredOnly GPT-5.6 Sol measured
OSWorld 2.027.3%Not measuredOnly GPT-5.6 Sol measured
BALROG60.0%Not measuredOnly GPT-5.6 Sol measured
PostTrainBench31.0%Not measuredOnly GPT-5.6 Sol measured
Vending-Bench 2$9,619Not measuredOnly GPT-5.6 Sol measured
LMArena Search1256Not measuredOnly GPT-5.6 Sol measured
MCP Atlas81.8%Not measuredOnly GPT-5.6 Sol measured
τ³-Banking46.9%Not measuredOnly GPT-5.6 Sol measured
Neither model has a result on the other 16 boards here
ReasoningDirectional onlyPublic boards, shown as published, kept out of the index20 boards, none shared
Reasoning benchmarks for GPT-5.6 Sol and Hy4 flash
BenchmarkGPT-5.6 SolHy4 flashLeader
GPQA Diamond95.2%Not measuredOnly GPT-5.6 Sol measured
MysteryMechanism (Vals AI)33.3%Not measuredOnly GPT-5.6 Sol measured
MMMU Pro (Vals AI)88.8%Not measuredOnly GPT-5.6 Sol measured
GPQA Diamond (Artificial Analysis run)94.1%Not measuredOnly GPT-5.6 Sol measured
Humanity's Last Exam (Artificial Analysis run)49.5%Not measuredOnly GPT-5.6 Sol measured
CritPt v2, hard subset (Artificial Analysis run)60.0%Not measuredOnly GPT-5.6 Sol measured
GPQA Diamond (Epoch AI run)93.5%Not measuredOnly GPT-5.6 Sol measured
ARC-AGI97.5%Not measuredOnly GPT-5.6 Sol measured
ARC-AGI-292.5%Not measuredOnly GPT-5.6 Sol measured
SimpleBench64.8%Not measuredOnly GPT-5.6 Sol measured
Chess puzzles59.0%Not measuredOnly GPT-5.6 Sol measured
Mystery game puzzles58.0%Not measuredOnly GPT-5.6 Sol measured
EBR-bench44.8%Not measuredOnly GPT-5.6 Sol measured
LMArena Text: hard prompts1483Not measuredOnly GPT-5.6 Sol measured
LMArena Text: expert1516Not measuredOnly GPT-5.6 Sol measured
LMArena Text: hard prompts in English1479Not measuredOnly GPT-5.6 Sol measured
LMArena Vision: diagrams1306Not measuredOnly GPT-5.6 Sol measured
LMArena Vision: homework1308Not measuredOnly GPT-5.6 Sol measured
Humanity's Last Exam (Scale run)31.2%Not measuredOnly GPT-5.6 Sol measured
GoBench1926Not measuredOnly GPT-5.6 Sol measured
Neither model has a result on the other 6 boards here
MathDirectional onlyPublic boards, shown as published, kept out of the index7 boards, none shared
Math benchmarks for GPT-5.6 Sol and Hy4 flash
BenchmarkGPT-5.6 SolHy4 flashLeader
OTIS Mock AIME100.0%Not measuredOnly GPT-5.6 Sol measured
FrontierMath, tiers 1 to 389.1%Not measuredOnly GPT-5.6 Sol measured
FrontierMath, tier 482.9%Not measuredOnly GPT-5.6 Sol measured
ProofBench83.0%Not measuredOnly GPT-5.6 Sol measured
LMArena Text: math1471Not measuredOnly GPT-5.6 Sol measured
LMArena Text: mathematical work1493Not measuredOnly GPT-5.6 Sol measured
MathArena56.3%Not measuredOnly GPT-5.6 Sol measured
Neither model has a result on the other 7 boards here
KnowledgeDirectional onlyPublic boards, shown as published, kept out of the index15 boards, none shared
Knowledge benchmarks for GPT-5.6 Sol and Hy4 flash
BenchmarkGPT-5.6 SolHy4 flashLeader
MMLU Pro89.1%Not measuredOnly GPT-5.6 Sol measured
LegalBench (Vals AI)87.0%Not measuredOnly GPT-5.6 Sol measured
CorpFin (Vals AI)64.4%Not measuredOnly GPT-5.6 Sol measured
TaxEval (Vals AI)74.8%Not measuredOnly GPT-5.6 Sol measured
MortgageTax (Vals AI)67.3%Not measuredOnly GPT-5.6 Sol measured
MedCode (Vals AI)44.0%Not measuredOnly GPT-5.6 Sol measured
SimpleQA Verified69.7%Not measuredOnly GPT-5.6 Sol measured
GeoBench73.0%Not measuredOnly GPT-5.6 Sol measured
LMArena Text: science1457Not measuredOnly GPT-5.6 Sol measured
LMArena Text: medicine and healthcare1429Not measuredOnly GPT-5.6 Sol measured
LMArena Text: legal and government1474Not measuredOnly GPT-5.6 Sol measured
LMArena Text: business and finance1457Not measuredOnly GPT-5.6 Sol measured
FACTS Benchmark Suite70.6%Not measuredOnly GPT-5.6 Sol measured
PRBench Finance50.4%Not measuredOnly GPT-5.6 Sol measured
PRBench Legal50.4%Not measuredOnly GPT-5.6 Sol measured
Neither model has a result on the other 6 boards here
Web designLike-for-likePublic boards, shown as published, kept out of the index50
Web design benchmarks for GPT-5.6 Sol and Hy4 flash
BenchmarkGPT-5.6 SolHy4 flashLeader
Design Arena: websites13141227GPT-5.6 Sol
Design Arena: UI components13211262GPT-5.6 Sol
Design Arena: game dev13501202GPT-5.6 Sol
Design Arena: data viz1319Not measuredOnly GPT-5.6 Sol measured
Design Arena: 3D13551209GPT-5.6 Sol
Design Arena: all code categories13311225GPT-5.6 Sol
Design Arena: SVG1298Not measuredOnly GPT-5.6 Sol measured
Design Arena: ASCII art1290Not measuredOnly GPT-5.6 Sol measured
Neither model has a result on the other 1 board here
GeneralDirectional onlyPublic boards, shown as published, kept out of the index36 boards, none shared
General benchmarks for GPT-5.6 Sol and Hy4 flash
BenchmarkGPT-5.6 SolHy4 flashLeader
Vals IndexNot measured50.8%Only Hy4 flash measured
Public Benefits Bench (Vals AI)66.5%Not measuredOnly GPT-5.6 Sol measured
MedScribe (Vals AI)85.2%Not measuredOnly GPT-5.6 Sol measured
Vals Multimodal Index72.6%Not measuredOnly GPT-5.6 Sol measured
IFBench72.7%Not measuredOnly GPT-5.6 Sol measured
Long context reasoning (AA-LCR)84.0%Not measuredOnly GPT-5.6 Sol measured
GDPval-AA1619Not measuredOnly GPT-5.6 Sol measured
DTBench95.7%Not measuredOnly GPT-5.6 Sol measured
LMCA58.5%Not measuredOnly GPT-5.6 Sol measured
Surface Evolver Bench93.1%Not measuredOnly GPT-5.6 Sol measured
Furniture assembly56.7%Not measuredOnly GPT-5.6 Sol measured
LMArena Text1457Not measuredOnly GPT-5.6 Sol measured
LMArena Text: instruction following1482Not measuredOnly GPT-5.6 Sol measured
LMArena Text: creative writing1451Not measuredOnly GPT-5.6 Sol measured
LMArena Vision1282Not measuredOnly GPT-5.6 Sol measured
LMArena Text: English1459Not measuredOnly GPT-5.6 Sol measured
LMArena Text: non-English1450Not measuredOnly GPT-5.6 Sol measured
LMArena Text: Chinese1526Not measuredOnly GPT-5.6 Sol measured
LMArena Text: German1472Not measuredOnly GPT-5.6 Sol measured
LMArena Text: Japanese1470Not measuredOnly GPT-5.6 Sol measured
LMArena Text: Korean1448Not measuredOnly GPT-5.6 Sol measured
LMArena Text: French1474Not measuredOnly GPT-5.6 Sol measured
LMArena Text: Spanish1444Not measuredOnly GPT-5.6 Sol measured
LMArena Text: Russian1471Not measuredOnly GPT-5.6 Sol measured
LMArena Text: Polish1463Not measuredOnly GPT-5.6 Sol measured
LMArena Text: multi-turn1462Not measuredOnly GPT-5.6 Sol measured
LMArena Text: longer queries1480Not measuredOnly GPT-5.6 Sol measured
LMArena Text: writing and language1464Not measuredOnly GPT-5.6 Sol measured
LMArena Text: entertainment and media1448Not measuredOnly GPT-5.6 Sol measured
LMArena Vision: English1284Not measuredOnly GPT-5.6 Sol measured
LMArena Vision: Chinese1309Not measuredOnly GPT-5.6 Sol measured
LMArena Vision: reading text1291Not measuredOnly GPT-5.6 Sol measured
LMArena Vision: humor1246Not measuredOnly GPT-5.6 Sol measured
LMArena Vision: creative writing1286Not measuredOnly GPT-5.6 Sol measured
LMArena Document1480Not measuredOnly GPT-5.6 Sol measured
EQ-Bench 41236Not measuredOnly GPT-5.6 Sol measured
Neither model has a result on the other 13 boards here
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
Image generation benchmarks for GPT-5.6 Sol and Hy4 flash
BenchmarkGPT-5.6 SolHy4 flashLeader
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

GPT-5.6 Sol$0.014Fits in one request
Hy4 flash$0.0001Fits in one request

Hy4 flash costs less

Repository review

50K input + 3K output tokens

GPT-5.6 Sol$0.260Fits in one request
Hy4 flash$0.0022Fits in one request

Hy4 flash costs less

Agent loop

200K input + 10K output tokens

GPT-5.6 Sol$1.00Fits in one request
Hy4 flash$0.0084Fits in one request

Hy4 flash 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.

Specifications of GPT-5.6 Sol and Hy4 flash
SpecificationOpenAI GPT-5.6 SolTencent Hy4 flash
MakerOpenAITencent
WeightsNo public weights listedOpen weights
ReleasedJuly 9, 2026November 16, 2026
Context window1.05M1.05M
Input $/Mtok$4.00$0.035
Output $/Mtok$20.00$0.1405
Coding index70.644.5
Independent measurements behind it52
Cost per solved task$1.20Not computable
SourceOpenRouter listingOpenRouter listing

Questions

Is GPT-5.6 Sol better than Hy4 flash for coding?

Pick GPT-5.6 Sol for coding: it scores 70.6 on the Vibe Coding Index against 44.5 for Hy4 flash. Pick Hy4 flash for output tokens 142.3x cheaper ($0.1405 vs $20.00 per million).

Which is cheaper, GPT-5.6 Sol or Hy4 flash?

GPT-5.6 Sol costs $4.00 per million input tokens and $20.00 per million output tokens; Hy4 flash costs $0.035 and $0.1405. Cost per solved task is known for GPT-5.6 Sol only.

Which has the bigger context window, GPT-5.6 Sol or Hy4 flash?

GPT-5.6 Sol, with 1.05M tokens against 1.05M.

How many benchmarks is the GPT-5.6 Sol vs Hy4 flash comparison based on?

10 benchmarks have a result for both models as of November 22, 2026. GPT-5.6 Sol's coding index rests on 5 independent benchmarks and Hy4 flash's on 2. Where only one model has a result, the page says so instead of naming a winner.

Compare next

Embed this comparison on your site

Free under CC BY 4.0, while the source link stays visible. Preview the frame · Terms and other embeds

<iframe src="https://llmbenchmark.ai/embed/compare/gpt-5-6-sol-vs-tencent-hy4-flash" title="GPT-5.6 Sol vs Hy4 flash, LLMBenchmark.ai" width="100%" height="360" style="border:0;max-width:100%" loading="lazy"></iframe>
<p style="font:13px/1.4 sans-serif;margin:6px 0 0">Source: <a href="https://llmbenchmark.ai/compare/gpt-5-6-sol-vs-tencent-hy4-flash">LLMBenchmark.ai, GPT-5.6 Sol vs Hy4 flash</a></p>