Kimi K2 Thinking vs Qwen3 Coder 480B A35B
Kimi K2 Thinking leads Qwen3 Coder 480B A35B on the Artificial Analysis Intelligence Index, 22.0 to 11.9, and Qwen3 Coder 480B A35B costs 1.7x less per output token ($1.50 vs $2.50 per million). Updated October 22, 2026; 16 benchmarks have a result for both.
Kimi K2 Thinking or Qwen3 Coder 480B A35B: which to pick
Neither model has a coding score yet, so the two cannot be ranked on coding.
Pick Kimi K2 Thinking for work measured by SWE-bench Verified or Design Arena: websites, where it leads.
- Shared results
- 16
- Kimi K2 Thinking only
- 14
- Qwen3 Coder 480B A35B only
- 37
- Like-for-like categories
- 8 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.
Coding
Vibe Coding Index, six public coding benchmarks
No call
Neither has a coding score yet.
One side not measured
Intelligence
AA Intelligence Index
Kimi K2 Thinking
Kimi K2 Thinking leads, 22.0 against 11.9.
Same board, both measured
Reasoning
GPQA Diamond
No call
Only Kimi K2 Thinking has a GPQA Diamond result, 78.5%.
Directional only
Math
AIME
No call
Only Kimi K2 Thinking has a AIME result, 85.4%.
Directional only
Knowledge
MMLU Pro
No call
Only Kimi K2 Thinking has a MMLU Pro result, 81.1%.
Directional only
Web design
Design Arena: websites
Qwen3 Coder 480B A35B
Qwen3 Coder 480B A35B leads, 1158 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
Qwen3 Coder 480B A35B
Kimi K2 Thinking $0.0019, Qwen3 Coder 480B A35B $0.0010 per turn.
List prices
Cost per solved task
One SWE-bench Verified task, solved
No call
Only Kimi K2 Thinking has the number, $0.61.
Directional only
At a glance
- Coding index
- Kimi K2 Thinking not benchmarked, Qwen3 Coder 480B A35B not benchmarked.
- Widest category
- Math, 55.4 points apart on average over 2 shared benchmarks.
- Widest single gap
- AA Math Index: Kimi K2 Thinking 94.7 against Qwen3 Coder 480B A35B 39.3, 55.4 points apart on a 0 to 100 scale.
- Output price
- Kimi K2 Thinking $2.50 against Qwen3 Coder 480B A35B $1.50 per million output tokens, 1.7x 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 Thinking | Bars | Qwen3 Coder 480B A35B | Ties |
|---|---|---|---|---|
| Coding | 5 | 1 | ||
| Intelligence | 1 | 0 | ||
| Agentic work | 1 | 0 | ||
| Reasoning | 2 | 0 | ||
| Math | 2 | 0 | ||
| Knowledge | 1 | 0 | ||
| Web design | 0 | 1 | ||
| General | 2 | 0 | ||
| All shared boards | 14 | 2 |
CodingLike-for-likeSix of these boards feed the Vibe Coding Index51
| Benchmark | Kimi K2 Thinking | Qwen3 Coder 480B A35B | Leader |
|---|---|---|---|
| SWE-bench Verified | 63.4% | 69.6% | Qwen3 Coder 480B A35B |
| LMArena WebDev | Not measured | 1275 | Only Qwen3 Coder 480B A35B measured |
| LiveCodeBench (Vals AI) | 63.1% | Not measured | Only Kimi K2 Thinking measured |
| SWE-bench Verified (Vals AI) | 60.2% | Not measured | Only Kimi K2 Thinking measured |
| LiveCodeBench (Artificial Analysis run) | 85.3% | 58.5% | Kimi K2 Thinking |
| Terminal-Bench Hard (Artificial Analysis run) | 31.1% | 18.9% | Kimi K2 Thinking |
| Terminal-Bench | 35.7% | 27.2% | Kimi K2 Thinking |
| GSO-Bench | Not measured | 4.9% | Only Qwen3 Coder 480B A35B measured |
| WeirdML | 42.8% | 41.2% | Kimi K2 Thinking |
| ALE-Bench | 597.5 | 461.4 | Kimi K2 Thinking |
| LMArena Text: coding | Not measured | 1412 | Only Qwen3 Coder 480B A35B measured |
| LMArena Text: software and IT | Not measured | 1393 | Only Qwen3 Coder 480B A35B measured |
| LMArena WebDev: React | Not measured | 1257 | Only Qwen3 Coder 480B A35B measured |
| LMArena WebDev: HTML | Not measured | 1287 | Only Qwen3 Coder 480B A35B measured |
| Neither model has a result on the other 25 boards here | |||
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index10
| Benchmark | Kimi K2 Thinking | Qwen3 Coder 480B A35B | Leader |
|---|---|---|---|
| AA Intelligence Index | 22.0 | 11.9 | Kimi K2 Thinking |
| Epoch Capabilities Index | 145.8 | Not measured | Only Kimi K2 Thinking measured |
Agentic workLike-for-likePublic boards, shown as published, kept out of the index10
| Benchmark | Kimi K2 Thinking | Qwen3 Coder 480B A35B | Leader |
|---|---|---|---|
| τ²-Bench Telecom | 93.0% | 43.6% | Kimi K2 Thinking |
| METR time horizons | 59.2% | Not measured | Only Kimi K2 Thinking measured |
| Neither model has a result on the other 39 boards here | |||
ReasoningLike-for-likePublic boards, shown as published, kept out of the index20
| Benchmark | Kimi K2 Thinking | Qwen3 Coder 480B A35B | Leader |
|---|---|---|---|
| GPQA Diamond | 78.5% | Not measured | Only Kimi K2 Thinking measured |
| GPQA Diamond (Artificial Analysis run) | 83.8% | 61.8% | Kimi K2 Thinking |
| Humanity's Last Exam (Artificial Analysis run) | 23.8% | 4.5% | Kimi K2 Thinking |
| LMArena Text: hard prompts | Not measured | 1372 | Only Qwen3 Coder 480B A35B measured |
| LMArena Text: expert | Not measured | 1338 | Only Qwen3 Coder 480B A35B measured |
| LMArena Text: hard prompts in English | Not measured | 1380 | Only Qwen3 Coder 480B A35B measured |
| Kagi LLM Benchmark | Not measured | 49.5% | Only Qwen3 Coder 480B A35B measured |
| Neither model has a result on the other 15 boards here | |||
MathLike-for-likePublic boards, shown as published, kept out of the index20
| Benchmark | Kimi K2 Thinking | Qwen3 Coder 480B A35B | Leader |
|---|---|---|---|
| AIME | 85.4% | Not measured | Only Kimi K2 Thinking measured |
| MGSM (Vals AI) | 90.2% | Not measured | Only Kimi K2 Thinking measured |
| AA Math Index | 94.7 | 39.3 | Kimi K2 Thinking |
| AIME 2025 (Artificial Analysis run) | 94.7% | 39.3% | Kimi K2 Thinking |
| LMArena Text: math | Not measured | 1365 | Only Qwen3 Coder 480B A35B measured |
| LMArena Text: mathematical work | Not measured | 1367 | Only Qwen3 Coder 480B A35B 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 | Kimi K2 Thinking | Qwen3 Coder 480B A35B | Leader |
|---|---|---|---|
| MMLU Pro | 81.1% | Not measured | Only Kimi K2 Thinking measured |
| LegalBench (Vals AI) | 80.2% | Not measured | Only Kimi K2 Thinking measured |
| CaseLaw (Vals AI) | 65.7% | Not measured | Only Kimi K2 Thinking measured |
| CorpFin (Vals AI) | 60.6% | Not measured | Only Kimi K2 Thinking measured |
| TaxEval (Vals AI) | 71.7% | Not measured | Only Kimi K2 Thinking measured |
| MedQA (Vals AI) | 92.6% | Not measured | Only Kimi K2 Thinking measured |
| MMLU Pro (Artificial Analysis run) | 84.8% | 78.8% | Kimi K2 Thinking |
| LMArena Text: science | Not measured | 1347 | Only Qwen3 Coder 480B A35B measured |
| LMArena Text: medicine and healthcare | Not measured | 1347 | Only Qwen3 Coder 480B A35B measured |
| LMArena Text: legal and government | Not measured | 1358 | Only Qwen3 Coder 480B A35B measured |
| LMArena Text: business and finance | Not measured | 1343 | Only Qwen3 Coder 480B A35B measured |
| Neither model has a result on the other 7 boards here | |||
Web designLike-for-likePublic boards, shown as published, kept out of the index01
| Benchmark | Kimi K2 Thinking | Qwen3 Coder 480B A35B | Leader |
|---|---|---|---|
| Design Arena: websites | 1113 | 1158 | Qwen3 Coder 480B A35B |
| Design Arena: UI components | Not measured | 1110 | Only Qwen3 Coder 480B A35B measured |
| Design Arena: game dev | Not measured | 1098 | Only Qwen3 Coder 480B A35B measured |
| Design Arena: data viz | Not measured | 1084 | Only Qwen3 Coder 480B A35B measured |
| Design Arena: all code categories | Not measured | 1142 | Only Qwen3 Coder 480B A35B measured |
| Neither model has a result on the other 4 boards here | |||
GeneralLike-for-likePublic boards, shown as published, kept out of the index20
| Benchmark | Kimi K2 Thinking | Qwen3 Coder 480B A35B | Leader |
|---|---|---|---|
| IFBench | 68.1% | 40.5% | Kimi K2 Thinking |
| Long context reasoning (AA-LCR) | 72.0% | 45.7% | Kimi K2 Thinking |
| CL-bench | 17.6% | Not measured | Only Kimi K2 Thinking measured |
| LMArena Text | Not measured | 1357 | Only Qwen3 Coder 480B A35B measured |
| LMArena Text: instruction following | Not measured | 1355 | Only Qwen3 Coder 480B A35B measured |
| LMArena Text: creative writing | Not measured | 1333 | Only Qwen3 Coder 480B A35B measured |
| LMArena Text: English | Not measured | 1359 | Only Qwen3 Coder 480B A35B measured |
| LMArena Text: non-English | Not measured | 1346 | Only Qwen3 Coder 480B A35B measured |
| LMArena Text: Chinese | Not measured | 1357 | Only Qwen3 Coder 480B A35B measured |
| LMArena Text: German | Not measured | 1325 | Only Qwen3 Coder 480B A35B measured |
| LMArena Text: Japanese | Not measured | 1311 | Only Qwen3 Coder 480B A35B measured |
| LMArena Text: Korean | Not measured | 1305 | Only Qwen3 Coder 480B A35B measured |
| LMArena Text: French | Not measured | 1398 | Only Qwen3 Coder 480B A35B measured |
| LMArena Text: Spanish | Not measured | 1360 | Only Qwen3 Coder 480B A35B measured |
| LMArena Text: Russian | Not measured | 1366 | Only Qwen3 Coder 480B A35B measured |
| LMArena Text: Polish | Not measured | 1344 | Only Qwen3 Coder 480B A35B measured |
| LMArena Text: multi-turn | Not measured | 1366 | Only Qwen3 Coder 480B A35B measured |
| LMArena Text: longer queries | Not measured | 1378 | Only Qwen3 Coder 480B A35B measured |
| LMArena Text: writing and language | Not measured | 1340 | Only Qwen3 Coder 480B A35B measured |
| LMArena Text: entertainment and media | Not measured | 1317 | Only Qwen3 Coder 480B A35B measured |
| Neither model has a result on the other 26 boards here | |||
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
| Benchmark | Kimi K2 Thinking | Qwen3 Coder 480B A35B | 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
Qwen3 Coder 480B A35B costs less
Repository review
50K input + 3K output tokens
Qwen3 Coder 480B A35B costs less
Agent loop
200K input + 10K output tokens
Qwen3 Coder 480B A35B 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 Thinking | Qwen3 Coder 480B A35B |
|---|---|---|
| Maker | Moonshot AI | Qwen |
| Weights | Open weights | Open weights |
| Released | November 6, 2025 | July 23, 2025 |
| Context window | 262K | 262K |
| Input $/Mtok | $0.60 | $0.30 |
| Output $/Mtok | $2.50 | $1.50 |
| Coding index | Not benchmarked | Not benchmarked |
| Independent measurements behind it | 1 | 1 |
| Cost per solved task | $0.61 | Not computable |
| Source | OpenRouter listing | OpenRouter listing |
Questions
Is Kimi K2 Thinking better than Qwen3 Coder 480B A35B for coding?
Neither model has a coding score yet, so the two cannot be ranked on coding. Pick Kimi K2 Thinking for work measured by SWE-bench Verified or Design Arena: websites, where it leads.
Which is cheaper, Kimi K2 Thinking or Qwen3 Coder 480B A35B?
Kimi K2 Thinking costs $0.60 per million input tokens and $2.50 per million output tokens; Qwen3 Coder 480B A35B costs $0.30 and $1.50. Cost per solved task is known for Kimi K2 Thinking only.
Which has the bigger context window, Kimi K2 Thinking or Qwen3 Coder 480B A35B?
Neither: both take 262K tokens.
How many benchmarks is the Kimi K2 Thinking vs Qwen3 Coder 480B A35B comparison based on?
16 benchmarks have a result for both models as of October 22, 2026. Kimi K2 Thinking's coding index rests on 1 independent benchmark and Qwen3 Coder 480B A35B's on 1. Where only one model has a result, the page says so instead of naming a winner.