Kimi K2.6 vs Hy3 preview
Updated October 21, 2026. 36 benchmarks have a result for both. Coding is ranked on the Vibe Coding Index; every other board is shown as published.
Decision reading
Pick Kimi K2.6 for coding: it scores 44.2 on the Vibe Coding Index against 22.4 for Hy3 preview.
Pick Hy3 preview when output tokens cost 5.7x less, or when the benchmark that matters is AA Agentic Index, where it leads.
- Shared results
- 36
- Kimi K2.6 only
- 73
- Hy3 preview only
- 0
- Like-for-like categories
- 7 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
Kimi K2.6
Kimi K2.6 scores 44.2 against 22.4.
One benchmark behind each score
Intelligence
AA Intelligence Index
Kimi K2.6
Kimi K2.6 leads, 27.0 against 25.3.
Same board, both measured
Agentic work
AA Agentic Index
Hy3 preview
Hy3 preview leads, 24.1 against 20.5.
Same board, both measured
Reasoning
GPQA Diamond
No call
Only Kimi K2.6 has a GPQA Diamond result, 89.1%.
Directional only
Knowledge
MMLU Pro
No call
Only Kimi K2.6 has a MMLU Pro result, 87.6%.
Directional only
Web design
Design Arena: websites
No call
Only Kimi K2.6 has a Design Arena: websites result, 1269.
Directional only
Long documents
Listed context window
No call
Both take 262K tokens.
Listed by OpenRouter
Everyday cost
1K input + 500 output tokens
Hy3 preview
Kimi K2.6 $0.0024, Hy3 preview $0.0005 per turn.
List prices
Cost per solved task
One SWE-bench Verified task, solved
No call
Only Kimi K2.6 has the number, $0.35.
Directional only
At a glance
- Coding index
- Kimi K2.6 44.2 from 1 benchmark, Hy3 preview 22.4 from 1 benchmark.
- Widest category
- Coding, 14.8 points apart on average over 7 shared benchmarks.
- Widest single gap
- LMArena WebDev: React: Kimi K2.6 1505 against Hy3 preview 1349, 26.1 points apart on a 0 to 100 scale.
- Output price
- Kimi K2.6 $3.41 against Hy3 preview $0.60 per million output tokens, 5.7x apart.
- Context window
- Both take 262K tokens.
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 | Kimi K2.6 | Bars | Hy3 preview | Ties |
|---|---|---|---|---|
| Coding | 7 | 0 | ||
| Intelligence | 1 | 0 | ||
| Agentic work | 1 | 1 | ||
| Reasoning | 5 | 0 | ||
| Math | 2 | 0 | ||
| Knowledge | 4 | 0 | ||
| General | 15 | 0 | ||
| All shared boards | 35 | 1 |
CodingLike-for-likeFour of these boards feed the Vibe Coding Index70
| Benchmark | Kimi K2.6 | Hy3 preview | Leader |
|---|---|---|---|
| LMArena WebDev | 1509 | 1357 | Kimi K2.6 |
| AA Coding Index | 61.8 | 58.8 | Kimi K2.6 |
| LiveCodeBench (Vals AI) | 86.8% | Not measured | Only Kimi K2.6 measured |
| SWE-bench Verified (Vals AI) | 76.2% | Not measured | Only Kimi K2.6 measured |
| Vibe Code Bench (Vals AI) | 37.9% | Not measured | Only Kimi K2.6 measured |
| Code migration (Vals AI) | 27.8% | Not measured | Only Kimi K2.6 measured |
| ProgramBench (Vals AI) | 0.0% | Not measured | Only Kimi K2.6 measured |
| Terminal-Bench 2.1 (Vals AI) | 53.6% | Not measured | Only Kimi K2.6 measured |
| SciCode | 51.5% | Not measured | Only Kimi K2.6 measured |
| Terminal-Bench Hard (Artificial Analysis run) | 43.9% | 34.1% | Kimi K2.6 |
| WeirdML | 55.9% | Not measured | Only Kimi K2.6 measured |
| ALE-Bench | 1092.7 | Not measured | Only Kimi K2.6 measured |
| LMArena Text: coding | 1489 | 1421 | Kimi K2.6 |
| LMArena Text: software and IT | 1484 | 1423 | Kimi K2.6 |
| LMArena WebDev: React | 1505 | 1349 | Kimi K2.6 |
| LMArena WebDev: HTML | 1485 | 1356 | Kimi K2.6 |
| LMArena WebDev: from an image | 1521 | Not measured | Only Kimi K2.6 measured |
| Neither model has a result on the other 21 boards here | |||
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index10
| Benchmark | Kimi K2.6 | Hy3 preview | Leader |
|---|---|---|---|
| AA Intelligence Index | 27.0 | 25.3 | Kimi K2.6 |
| Epoch Capabilities Index | 150.9 | Not measured | Only Kimi K2.6 measured |
Agentic workLike-for-likePublic boards, shown as published, kept out of the index11
| Benchmark | Kimi K2.6 | Hy3 preview | Leader |
|---|---|---|---|
| AA Agentic Index | 20.5 | 24.1 | Hy3 preview |
| Design Arena: agents building web apps | 1268 | Not measured | Only Kimi K2.6 measured |
| Design Arena: agents building full-stack apps | 1155 | Not measured | Only Kimi K2.6 measured |
| Design Arena: agents building mobile apps | 1156 | Not measured | Only Kimi K2.6 measured |
| Design Arena: agents building Android apps | 1166 | Not measured | Only Kimi K2.6 measured |
| Design Arena: agents building Godot games | 1120 | Not measured | Only Kimi K2.6 measured |
| Design Arena: agents building games | 1141 | Not measured | Only Kimi K2.6 measured |
| Design Arena: HTML slides | 1184 | Not measured | Only Kimi K2.6 measured |
| Design Arena: PowerPoint slides | 1128 | Not measured | Only Kimi K2.6 measured |
| Design Arena: spreadsheets | 1276 | Not measured | Only Kimi K2.6 measured |
| Finance Agent (Vals AI) | 44.9% | Not measured | Only Kimi K2.6 measured |
| Tax Agent Bench (Vals AI) | 12.4% | Not measured | Only Kimi K2.6 measured |
| Legal Research Bench (Vals AI) | 15.9% | Not measured | Only Kimi K2.6 measured |
| Excel Modeling Benchmark (Vals AI) | 57.9% | Not measured | Only Kimi K2.6 measured |
| τ²-Bench Telecom | 95.9% | 92.7% | Kimi K2.6 |
| OSWorld 2.0 | 4.6% | Not measured | Only Kimi K2.6 measured |
| Vending-Bench 2 | $6,205 | Not measured | Only Kimi K2.6 measured |
| Neither model has a result on the other 22 boards here | |||
ReasoningLike-for-likePublic boards, shown as published, kept out of the index50
| Benchmark | Kimi K2.6 | Hy3 preview | Leader |
|---|---|---|---|
| GPQA Diamond | 89.1% | Not measured | Only Kimi K2.6 measured |
| MMMU Pro (Vals AI) | 86.3% | Not measured | Only Kimi K2.6 measured |
| GPQA Diamond (Artificial Analysis run) | 91.1% | 86.7% | Kimi K2.6 |
| Humanity's Last Exam (Artificial Analysis run) | 37.5% | 27.8% | Kimi K2.6 |
| GPQA Diamond (Epoch AI run) | 90.8% | Not measured | Only Kimi K2.6 measured |
| Chess puzzles | 20.0% | Not measured | Only Kimi K2.6 measured |
| Mystery game puzzles | 18.0% | Not measured | Only Kimi K2.6 measured |
| EBR-bench | 2.4% | Not measured | Only Kimi K2.6 measured |
| LMArena Text: hard prompts | 1470 | 1413 | Kimi K2.6 |
| LMArena Text: expert | 1491 | 1422 | Kimi K2.6 |
| LMArena Text: hard prompts in English | 1471 | 1416 | Kimi K2.6 |
| LMArena Vision: diagrams | 1298 | Not measured | Only Kimi K2.6 measured |
| LMArena Vision: homework | 1310 | Not measured | Only Kimi K2.6 measured |
| Neither model has a result on the other 8 boards here | |||
MathLike-for-likePublic boards, shown as published, kept out of the index20
| Benchmark | Kimi K2.6 | Hy3 preview | Leader |
|---|---|---|---|
| OTIS Mock AIME | 96.1% | Not measured | Only Kimi K2.6 measured |
| FrontierMath, tiers 1 to 3 | 57.2% | Not measured | Only Kimi K2.6 measured |
| FrontierMath, tier 4 | 25.6% | Not measured | Only Kimi K2.6 measured |
| ProofBench | 16.0% | Not measured | Only Kimi K2.6 measured |
| LMArena Text: math | 1474 | 1424 | Kimi K2.6 |
| LMArena Text: mathematical work | 1480 | 1442 | Kimi K2.6 |
| Neither model has a result on the other 7 boards here | |||
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index40
| Benchmark | Kimi K2.6 | Hy3 preview | Leader |
|---|---|---|---|
| MMLU Pro | 87.6% | Not measured | Only Kimi K2.6 measured |
| LegalBench (Vals AI) | 84.7% | Not measured | Only Kimi K2.6 measured |
| CaseLaw (Vals AI) | 61.2% | Not measured | Only Kimi K2.6 measured |
| CorpFin (Vals AI) | 66.7% | Not measured | Only Kimi K2.6 measured |
| TaxEval (Vals AI) | 74.7% | Not measured | Only Kimi K2.6 measured |
| MortgageTax (Vals AI) | 65.8% | Not measured | Only Kimi K2.6 measured |
| MedCode (Vals AI) | 40.1% | Not measured | Only Kimi K2.6 measured |
| SimpleQA Verified | 34.9% | Not measured | Only Kimi K2.6 measured |
| LMArena Text: science | 1468 | 1416 | Kimi K2.6 |
| LMArena Text: medicine and healthcare | 1457 | 1405 | Kimi K2.6 |
| LMArena Text: legal and government | 1462 | 1430 | Kimi K2.6 |
| LMArena Text: business and finance | 1445 | 1397 | Kimi K2.6 |
| Neither model has a result on the other 5 boards here | |||
Web designDirectional onlyPublic boards, shown as published, kept out of the index8 boards, none shared
| Benchmark | Kimi K2.6 | Hy3 preview | Leader |
|---|---|---|---|
| Design Arena: websites | 1269 | Not measured | Only Kimi K2.6 measured |
| Design Arena: UI components | 1266 | Not measured | Only Kimi K2.6 measured |
| Design Arena: game dev | 1251 | Not measured | Only Kimi K2.6 measured |
| Design Arena: data viz | 1249 | Not measured | Only Kimi K2.6 measured |
| Design Arena: 3D | 1268 | Not measured | Only Kimi K2.6 measured |
| Design Arena: all code categories | 1272 | Not measured | Only Kimi K2.6 measured |
| Design Arena: SVG | 1181 | Not measured | Only Kimi K2.6 measured |
| Design Arena: ASCII art | 1164 | Not measured | Only Kimi K2.6 measured |
| Neither model has a result on the other 1 board here | |||
GeneralLike-for-likePublic boards, shown as published, kept out of the index150
| Benchmark | Kimi K2.6 | Hy3 preview | Leader |
|---|---|---|---|
| Public Benefits Bench (Vals AI) | 56.6% | Not measured | Only Kimi K2.6 measured |
| SAGE (Vals AI) | 50.2% | Not measured | Only Kimi K2.6 measured |
| MedScribe (Vals AI) | 78.2% | Not measured | Only Kimi K2.6 measured |
| Vals Multimodal Index | 56.4% | Not measured | Only Kimi K2.6 measured |
| IFBench | 76.0% | 63.1% | Kimi K2.6 |
| Long context reasoning (AA-LCR) | 81.0% | 64.7% | Kimi K2.6 |
| DTBench | 90.9% | Not measured | Only Kimi K2.6 measured |
| LMCA | 37.3% | Not measured | Only Kimi K2.6 measured |
| Furniture assembly | 21.7% | Not measured | Only Kimi K2.6 measured |
| ExploitBench | 18.4% | Not measured | Only Kimi K2.6 measured |
| LMArena Text | 1455 | 1401 | Kimi K2.6 |
| LMArena Text: instruction following | 1451 | 1383 | Kimi K2.6 |
| LMArena Text: creative writing | 1434 | 1339 | Kimi K2.6 |
| LMArena Vision | 1282 | Not measured | Only Kimi K2.6 measured |
| LMArena Text: English | 1460 | 1409 | Kimi K2.6 |
| LMArena Text: non-English | 1446 | 1390 | Kimi K2.6 |
| LMArena Text: Chinese | 1520 | 1435 | Kimi K2.6 |
| LMArena Text: German | 1451 | Not measured | Only Kimi K2.6 measured |
| LMArena Text: Japanese | 1443 | Not measured | Only Kimi K2.6 measured |
| LMArena Text: Korean | 1427 | Not measured | Only Kimi K2.6 measured |
| LMArena Text: French | 1471 | 1425 | Kimi K2.6 |
| LMArena Text: Spanish | 1464 | 1399 | Kimi K2.6 |
| LMArena Text: Russian | 1446 | 1399 | Kimi K2.6 |
| LMArena Text: Polish | 1438 | Not measured | Only Kimi K2.6 measured |
| LMArena Text: multi-turn | 1453 | 1411 | Kimi K2.6 |
| LMArena Text: longer queries | 1468 | 1402 | Kimi K2.6 |
| LMArena Text: writing and language | 1439 | 1370 | Kimi K2.6 |
| LMArena Text: entertainment and media | 1428 | 1352 | Kimi K2.6 |
| LMArena Vision: English | 1287 | Not measured | Only Kimi K2.6 measured |
| LMArena Vision: Chinese | 1346 | Not measured | Only Kimi K2.6 measured |
| LMArena Vision: reading text | 1291 | Not measured | Only Kimi K2.6 measured |
| LMArena Vision: humor | 1280 | Not measured | Only Kimi K2.6 measured |
| LMArena Vision: creative writing | 1273 | Not measured | Only Kimi K2.6 measured |
| LMArena Document | 1451 | Not measured | Only Kimi K2.6 measured |
| Neither model has a result on the other 12 boards here | |||
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
| Benchmark | Kimi K2.6 | Hy3 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
Hy3 preview costs less
Repository review
50K input + 3K output tokens
Hy3 preview costs less
Agent loop
200K input + 10K output tokens
Hy3 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 | MoonshotAI Kimi K2.6 | Tencent Hy3 preview |
|---|---|---|
| Maker | Moonshot AI | Tencent |
| Weights | Open weights | Open weights |
| Released | April 20, 2026 | April 22, 2026 |
| Context window | 262K | 262K |
| Input $/Mtok | $0.65 | $0.18 |
| Output $/Mtok | $3.41 | $0.60 |
| Coding index | 44.2 | 22.4 |
| Independent measurements behind it | 1 | 1 |
| Cost per solved task | $0.35 | Not computable |
| Source | OpenRouter listing | OpenRouter listing |
Questions
Is Kimi K2.6 better than Hy3 preview for coding?
Pick Kimi K2.6 for coding: it scores 44.2 on the Vibe Coding Index against 22.4 for Hy3 preview. Pick Hy3 preview when output tokens cost 5.7x less, or when the benchmark that matters is AA Agentic Index, where it leads.
Which is cheaper, Kimi K2.6 or Hy3 preview?
Kimi K2.6 costs $0.65 per million input tokens and $3.41 per million output tokens; Hy3 preview costs $0.18 and $0.60. Cost per solved task is known for Kimi K2.6 only.
Which has the bigger context window, Kimi K2.6 or Hy3 preview?
Neither: both take 262K tokens.
How many benchmarks is the Kimi K2.6 vs Hy3 preview comparison based on?
36 benchmarks have a result for both models as of October 21, 2026. Kimi K2.6's coding index rests on 1 independent benchmark and Hy3 preview's on 1. Where only one model has a result, the page says so instead of naming a winner.