Kimi K2.6 vs Kimi K2 Thinking
Updated October 21, 2026. 19 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 Thinking for coding: it scores 63.4 on the Vibe Coding Index against 44.2 for Kimi K2.6.
Pick Kimi K2.6 when a solved task costs $0.35 against $0.61, or when what matters is LiveCodeBench (Vals AI) or SWE-bench Verified (Vals AI), where it leads.
- Shared results
- 19
- Kimi K2.6 only
- 90
- Kimi K2 Thinking only
- 11
- 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 Thinking
Kimi K2 Thinking scores 63.4 against 44.2.
One benchmark behind each score
Intelligence
AA Intelligence Index
Kimi K2.6
Kimi K2.6 leads, 27.0 against 22.0.
Same board, both measured
Agentic work
AA Agentic Index
No call
Only Kimi K2.6 has a AA Agentic Index result, 20.5.
Directional only
Reasoning
GPQA Diamond
Kimi K2.6
Kimi K2.6 leads, 89.1% against 78.5%.
Same board, both measured
Math
AIME
No call
Only Kimi K2 Thinking has a AIME result, 85.4%.
Directional only
Knowledge
MMLU Pro
Kimi K2.6
Kimi K2.6 leads, 87.6% against 81.1%.
Same board, both measured
Web design
Design Arena: websites
Kimi K2.6
Kimi K2.6 leads, 1269 against 1113.
Same board, both measured
Long documents
Listed context window
No call
Both take 262K tokens.
Listed by OpenRouter
Everyday cost
1K input + 500 output tokens
Kimi K2 Thinking
Kimi K2.6 $0.0024, Kimi K2 Thinking $0.0019 per turn.
List prices
Cost per solved task
One SWE-bench Verified task, solved
Kimi K2.6
Kimi K2.6 $0.35, Kimi K2 Thinking $0.61.
Published attempt cost and pass rate
At a glance
- Coding index
- Kimi K2.6 44.2 from 1 benchmark, Kimi K2 Thinking 63.4 from 1 benchmark.
- Widest category
- Web design, 19.5 points apart on average over 1 shared benchmark.
- Widest single gap
- LiveCodeBench (Vals AI): Kimi K2.6 86.8% against Kimi K2 Thinking 63.1%, 23.6 points apart on a 0 to 100 scale.
- Output price
- Kimi K2.6 $3.41 against Kimi K2 Thinking $2.50 per million output tokens, 1.4x 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 | Kimi K2 Thinking | Ties |
|---|---|---|---|---|
| Coding | 5 | 0 | ||
| Intelligence | 2 | 0 | ||
| Agentic work | 1 | 0 | ||
| Reasoning | 3 | 0 | ||
| Knowledge | 4 | 1 | ||
| Web design | 1 | 0 | ||
| General | 2 | 0 | ||
| All shared boards | 18 | 1 |
CodingLike-for-likeFour of these boards feed the Vibe Coding Index50
| Benchmark | Kimi K2.6 | Kimi K2 Thinking | Leader |
|---|---|---|---|
| SWE-bench Verified | Not measured | 63.4% | Only Kimi K2 Thinking measured |
| LMArena WebDev | 1509 | Not measured | Only Kimi K2.6 measured |
| AA Coding Index | 61.8 | Not measured | Only Kimi K2.6 measured |
| LiveCodeBench (Vals AI) | 86.8% | 63.1% | Kimi K2.6 |
| SWE-bench Verified (Vals AI) | 76.2% | 60.2% | Kimi K2.6 |
| 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 |
| LiveCodeBench (Artificial Analysis run) | Not measured | 85.3% | Only Kimi K2 Thinking measured |
| SciCode | 51.5% | Not measured | Only Kimi K2.6 measured |
| Terminal-Bench Hard (Artificial Analysis run) | 43.9% | 31.1% | Kimi K2.6 |
| Terminal-Bench | Not measured | 35.7% | Only Kimi K2 Thinking measured |
| WeirdML | 55.9% | 42.8% | Kimi K2.6 |
| ALE-Bench | 1092.7 | 597.5 | Kimi K2.6 |
| LMArena Text: coding | 1489 | Not measured | Only Kimi K2.6 measured |
| LMArena Text: software and IT | 1484 | Not measured | Only Kimi K2.6 measured |
| LMArena WebDev: React | 1505 | Not measured | Only Kimi K2.6 measured |
| LMArena WebDev: HTML | 1485 | Not measured | Only Kimi K2.6 measured |
| LMArena WebDev: from an image | 1521 | Not measured | Only Kimi K2.6 measured |
| Neither model has a result on the other 18 boards here | |||
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index20
| Benchmark | Kimi K2.6 | Kimi K2 Thinking | Leader |
|---|---|---|---|
| AA Intelligence Index | 27.0 | 22.0 | Kimi K2.6 |
| Epoch Capabilities Index | 150.9 | 145.8 | Kimi K2.6 |
Agentic workLike-for-likePublic boards, shown as published, kept out of the index10
| Benchmark | Kimi K2.6 | Kimi K2 Thinking | Leader |
|---|---|---|---|
| AA Agentic Index | 20.5 | Not measured | Only Kimi K2.6 measured |
| 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% | 93.0% | Kimi K2.6 |
| METR time horizons | Not measured | 59.2% | Only Kimi K2 Thinking measured |
| 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 21 boards here | |||
ReasoningLike-for-likePublic boards, shown as published, kept out of the index30
| Benchmark | Kimi K2.6 | Kimi K2 Thinking | Leader |
|---|---|---|---|
| GPQA Diamond | 89.1% | 78.5% | Kimi K2.6 |
| MMMU Pro (Vals AI) | 86.3% | Not measured | Only Kimi K2.6 measured |
| GPQA Diamond (Artificial Analysis run) | 91.1% | 83.8% | Kimi K2.6 |
| Humanity's Last Exam (Artificial Analysis run) | 37.5% | 23.8% | 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 | Not measured | Only Kimi K2.6 measured |
| LMArena Text: expert | 1491 | Not measured | Only Kimi K2.6 measured |
| LMArena Text: hard prompts in English | 1471 | Not measured | Only Kimi K2.6 measured |
| 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 | |||
MathDirectional onlyPublic boards, shown as published, kept out of the index10 boards, none shared
| Benchmark | Kimi K2.6 | Kimi K2 Thinking | Leader |
|---|---|---|---|
| AIME | Not measured | 85.4% | Only Kimi K2 Thinking measured |
| MGSM (Vals AI) | Not measured | 90.2% | Only Kimi K2 Thinking measured |
| AA Math Index | Not measured | 94.7 | Only Kimi K2 Thinking measured |
| AIME 2025 (Artificial Analysis run) | Not measured | 94.7% | Only Kimi K2 Thinking measured |
| 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 | Not measured | Only Kimi K2.6 measured |
| LMArena Text: mathematical work | 1480 | Not measured | Only Kimi K2.6 measured |
| Neither model has a result on the other 3 boards here | |||
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index41
| Benchmark | Kimi K2.6 | Kimi K2 Thinking | Leader |
|---|---|---|---|
| MMLU Pro | 87.6% | 81.1% | Kimi K2.6 |
| LegalBench (Vals AI) | 84.7% | 80.2% | Kimi K2.6 |
| CaseLaw (Vals AI) | 61.2% | 65.7% | Kimi K2 Thinking |
| CorpFin (Vals AI) | 66.7% | 60.6% | Kimi K2.6 |
| TaxEval (Vals AI) | 74.7% | 71.7% | Kimi K2.6 |
| MortgageTax (Vals AI) | 65.8% | Not measured | Only Kimi K2.6 measured |
| MedQA (Vals AI) | Not measured | 92.6% | Only Kimi K2 Thinking measured |
| MedCode (Vals AI) | 40.1% | Not measured | Only Kimi K2.6 measured |
| MMLU Pro (Artificial Analysis run) | Not measured | 84.8% | Only Kimi K2 Thinking measured |
| SimpleQA Verified | 34.9% | Not measured | Only Kimi K2.6 measured |
| LMArena Text: science | 1468 | Not measured | Only Kimi K2.6 measured |
| LMArena Text: medicine and healthcare | 1457 | Not measured | Only Kimi K2.6 measured |
| LMArena Text: legal and government | 1462 | Not measured | Only Kimi K2.6 measured |
| LMArena Text: business and finance | 1445 | Not measured | Only Kimi K2.6 measured |
| Neither model has a result on the other 3 boards here | |||
Web designLike-for-likePublic boards, shown as published, kept out of the index10
| Benchmark | Kimi K2.6 | Kimi K2 Thinking | Leader |
|---|---|---|---|
| Design Arena: websites | 1269 | 1113 | Kimi K2.6 |
| 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 index20
| Benchmark | Kimi K2.6 | Kimi K2 Thinking | 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% | 68.1% | Kimi K2.6 |
| Long context reasoning (AA-LCR) | 81.0% | 72.0% | Kimi K2.6 |
| DTBench | 90.9% | Not measured | Only Kimi K2.6 measured |
| LMCA | 37.3% | Not measured | Only Kimi K2.6 measured |
| CL-bench | Not measured | 17.6% | Only Kimi K2 Thinking 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 | Not measured | Only Kimi K2.6 measured |
| LMArena Text: instruction following | 1451 | Not measured | Only Kimi K2.6 measured |
| LMArena Text: creative writing | 1434 | Not measured | Only Kimi K2.6 measured |
| LMArena Vision | 1282 | Not measured | Only Kimi K2.6 measured |
| LMArena Text: English | 1460 | Not measured | Only Kimi K2.6 measured |
| LMArena Text: non-English | 1446 | Not measured | Only Kimi K2.6 measured |
| LMArena Text: Chinese | 1520 | Not measured | Only Kimi K2.6 measured |
| 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 | Not measured | Only Kimi K2.6 measured |
| LMArena Text: Spanish | 1464 | Not measured | Only Kimi K2.6 measured |
| LMArena Text: Russian | 1446 | Not measured | Only Kimi K2.6 measured |
| LMArena Text: Polish | 1438 | Not measured | Only Kimi K2.6 measured |
| LMArena Text: multi-turn | 1453 | Not measured | Only Kimi K2.6 measured |
| LMArena Text: longer queries | 1468 | Not measured | Only Kimi K2.6 measured |
| LMArena Text: writing and language | 1439 | Not measured | Only Kimi K2.6 measured |
| LMArena Text: entertainment and media | 1428 | Not measured | Only Kimi K2.6 measured |
| 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 11 boards here | |||
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
| Benchmark | Kimi K2.6 | Kimi K2 Thinking | 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
Kimi K2 Thinking costs less
Repository review
50K input + 3K output tokens
Kimi K2 Thinking costs less
Agent loop
200K input + 10K output tokens
Kimi K2 Thinking 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 | MoonshotAI Kimi K2 Thinking |
|---|---|---|
| Maker | Moonshot AI | Moonshot AI |
| Weights | Open weights | Open weights |
| Released | April 20, 2026 | November 6, 2025 |
| Context window | 262K | 262K |
| Input $/Mtok | $0.65 | $0.60 |
| Output $/Mtok | $3.41 | $2.50 |
| Coding index | 44.2 | 63.4 |
| Independent measurements behind it | 1 | 1 |
| Cost per solved task | $0.35 | $0.61 |
| Source | OpenRouter listing | OpenRouter listing |
Questions
Is Kimi K2.6 better than Kimi K2 Thinking for coding?
Pick Kimi K2 Thinking for coding: it scores 63.4 on the Vibe Coding Index against 44.2 for Kimi K2.6. Pick Kimi K2.6 when a solved task costs $0.35 against $0.61, or when what matters is LiveCodeBench (Vals AI) or SWE-bench Verified (Vals AI), where it leads.
Which is cheaper, Kimi K2.6 or Kimi K2 Thinking?
Kimi K2.6 costs $0.65 per million input tokens and $3.41 per million output tokens; Kimi K2 Thinking costs $0.60 and $2.50. Per solved SWE-bench Verified task, Kimi K2.6 costs $0.35 and Kimi K2 Thinking $0.61.
Which has the bigger context window, Kimi K2.6 or Kimi K2 Thinking?
Neither: both take 262K tokens.
How many benchmarks is the Kimi K2.6 vs Kimi K2 Thinking comparison based on?
19 benchmarks have a result for both models as of October 21, 2026. Kimi K2.6's coding index rests on 1 independent benchmark and Kimi K2 Thinking's on 1. Where only one model has a result, the page says so instead of naming a winner.