Qwen3 30B A3B vs Hy3
Qwen3 30B A3B and Hy3 cost about the same per output token ($0.50 vs $0.528 per million). Updated October 22, 2026; 36 benchmarks have a result for both.
Qwen3 30B A3B or Hy3: which to pick
Neither model has a coding score yet, so the two cannot be ranked on coding.
Pick Qwen3 30B A3B for output tokens 1.1x cheaper ($0.50 vs $0.528 per million) and work measured by LMArena Text: coding or LMArena Text: software and IT, where it leads.
- Shared results
- 36
- Qwen3 30B A3B only
- 16
- Hy3 only
- 8
- Like-for-like categories
- 6 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
No call
Only Qwen3 30B A3B has a AA Intelligence Index result, 7.6.
Directional only
Web design
Design Arena: websites
Hy3
Hy3 leads, 1230 against 954.
Same board, both measured
Long documents
Listed context window
Hy3
Hy3 takes 262K tokens against 41K.
Listed by OpenRouter
Everyday cost
1K input + 500 output tokens
Qwen3 30B A3B
Qwen3 30B A3B $0.0004, Hy3 $0.0004 per turn.
List prices
At a glance
- Coding index
- Qwen3 30B A3B not benchmarked, Hy3 not benchmarked.
- Widest category
- Web design, 33.3 points apart on average over 5 shared benchmarks.
- Widest single gap
- Long context reasoning (AA-LCR): Qwen3 30B A3B 0.0% against Hy3 79.0%, 79.0 points apart on a 0 to 100 scale.
- Output price
- Qwen3 30B A3B $0.50 against Hy3 $0.528 per million output tokens, 1.1x apart.
- Context window
- Qwen3 30B A3B takes 41K tokens, Hy3 262K.
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 | Qwen3 30B A3B | Bars | Hy3 | Ties |
|---|---|---|---|---|
| Coding | 0 | 2 | ||
| Reasoning | 0 | 5 | ||
| Math | 0 | 2 | ||
| Knowledge | 0 | 4 | ||
| Web design | 0 | 5 | ||
| General | 0 | 18 | ||
| All shared boards | 0 | 36 |
CodingLike-for-likeSix of these boards feed the Vibe Coding Index02
| Benchmark | Qwen3 30B A3B | Hy3 | Leader |
|---|---|---|---|
| LMArena WebDev | Not measured | 1507 | Only Hy3 measured |
| LiveCodeBench (Artificial Analysis run) | 50.6% | Not measured | Only Qwen3 30B A3B measured |
| SciCode | Not measured | 48.6% | Only Hy3 measured |
| Terminal-Bench Hard (Artificial Analysis run) | 6.8% | Not measured | Only Qwen3 30B A3B measured |
| WeirdML | 29.8% | Not measured | Only Qwen3 30B A3B measured |
| ALE-Bench | Not measured | 838.6 | Only Hy3 measured |
| LMArena Text: coding | 1337 | 1466 | Hy3 |
| LMArena Text: software and IT | 1340 | 1463 | Hy3 |
| LMArena WebDev: React | Not measured | 1511 | Only Hy3 measured |
| LMArena WebDev: HTML | Not measured | 1487 | Only Hy3 measured |
| Neither model has a result on the other 29 boards here | |||
IntelligenceDirectional onlyPublic boards, shown as published, kept out of the index2 boards, none shared
| Benchmark | Qwen3 30B A3B | Hy3 | Leader |
|---|---|---|---|
| AA Intelligence Index | 7.6 | Not measured | Only Qwen3 30B A3B measured |
| Epoch Capabilities Index | 137.3 | Not measured | Only Qwen3 30B A3B measured |
Agentic workDirectional onlyPublic boards, shown as published, kept out of the index1 board, none shared
| Benchmark | Qwen3 30B A3B | Hy3 | Leader |
|---|---|---|---|
| τ²-Bench Telecom | 26.0% | Not measured | Only Qwen3 30B A3B measured |
| Neither model has a result on the other 40 boards here | |||
ReasoningLike-for-likePublic boards, shown as published, kept out of the index05
| Benchmark | Qwen3 30B A3B | Hy3 | Leader |
|---|---|---|---|
| GPQA Diamond (Artificial Analysis run) | 61.6% | 89.7% | Hy3 |
| Humanity's Last Exam (Artificial Analysis run) | 6.2% | 33.5% | Hy3 |
| GPQA Diamond (Epoch AI run) | 61.7% | Not measured | Only Qwen3 30B A3B measured |
| LMArena Text: hard prompts | 1314 | 1447 | Hy3 |
| LMArena Text: expert | 1314 | 1461 | Hy3 |
| LMArena Text: hard prompts in English | 1329 | 1455 | Hy3 |
| Neither model has a result on the other 16 boards here | |||
MathLike-for-likePublic boards, shown as published, kept out of the index02
| Benchmark | Qwen3 30B A3B | Hy3 | Leader |
|---|---|---|---|
| AA Math Index | 72.3 | Not measured | Only Qwen3 30B A3B measured |
| AIME 2025 (Artificial Analysis run) | 72.3% | Not measured | Only Qwen3 30B A3B measured |
| OTIS Mock AIME | 62.8% | Not measured | Only Qwen3 30B A3B measured |
| LMArena Text: math | 1355 | 1473 | Hy3 |
| LMArena Text: mathematical work | 1363 | 1482 | Hy3 |
| Neither model has a result on the other 8 boards here | |||
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index04
| Benchmark | Qwen3 30B A3B | Hy3 | Leader |
|---|---|---|---|
| MMLU Pro (Artificial Analysis run) | 77.7% | Not measured | Only Qwen3 30B A3B measured |
| LMArena Text: science | 1321 | 1457 | Hy3 |
| LMArena Text: medicine and healthcare | 1318 | 1452 | Hy3 |
| LMArena Text: legal and government | 1314 | 1422 | Hy3 |
| LMArena Text: business and finance | 1309 | 1424 | Hy3 |
| Neither model has a result on the other 13 boards here | |||
Web designLike-for-likePublic boards, shown as published, kept out of the index05
| Benchmark | Qwen3 30B A3B | Hy3 | Leader |
|---|---|---|---|
| Design Arena: websites | 954 | 1230 | Hy3 |
| Design Arena: UI components | 943 | 1204 | Hy3 |
| Design Arena: game dev | 898 | 1208 | Hy3 |
| Design Arena: data viz | 967 | 1164 | Hy3 |
| Design Arena: 3D | Not measured | 1232 | Only Hy3 measured |
| Design Arena: all code categories | 941 | 1227 | Hy3 |
| Design Arena: SVG | Not measured | 1200 | Only Hy3 measured |
| Design Arena: ASCII art | Not measured | 1152 | Only Hy3 measured |
| Neither model has a result on the other 1 board here | |||
GeneralLike-for-likePublic boards, shown as published, kept out of the index018
| Benchmark | Qwen3 30B A3B | Hy3 | Leader |
|---|---|---|---|
| IFBench | 41.5% | Not measured | Only Qwen3 30B A3B measured |
| Long context reasoning (AA-LCR) | 0.0% | 79.0% | Hy3 |
| DTBench | 60.3% | Not measured | Only Qwen3 30B A3B measured |
| LMCA | 15.8% | Not measured | Only Qwen3 30B A3B measured |
| Fiction.LiveBench | 40.6% | Not measured | Only Qwen3 30B A3B measured |
| Creative writing (Lech Mazur) | 75.3% | Not measured | Only Qwen3 30B A3B measured |
| LMArena Text | 1317 | 1438 | Hy3 |
| LMArena Text: instruction following | 1283 | 1426 | Hy3 |
| LMArena Text: creative writing | 1271 | 1403 | Hy3 |
| LMArena Text: English | 1335 | 1448 | Hy3 |
| LMArena Text: non-English | 1295 | 1426 | Hy3 |
| LMArena Text: Chinese | 1347 | 1485 | Hy3 |
| LMArena Text: German | 1307 | 1449 | Hy3 |
| LMArena Text: Japanese | 1254 | 1397 | Hy3 |
| LMArena Text: Korean | 1261 | 1396 | Hy3 |
| LMArena Text: French | 1352 | 1456 | Hy3 |
| LMArena Text: Spanish | 1317 | 1457 | Hy3 |
| LMArena Text: Russian | 1291 | 1432 | Hy3 |
| LMArena Text: Polish | 1299 | 1447 | Hy3 |
| LMArena Text: multi-turn | 1307 | 1436 | Hy3 |
| LMArena Text: longer queries | 1312 | 1442 | Hy3 |
| LMArena Text: writing and language | 1290 | 1414 | Hy3 |
| LMArena Text: entertainment and media | 1258 | 1400 | Hy3 |
| Neither model has a result on the other 23 boards here | |||
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
| Benchmark | Qwen3 30B A3B | Hy3 | 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 30B A3B costs less
Repository review
50K input + 3K output tokens
Qwen3 30B A3B costs less
Agent loop
200K input + 10K output tokens
Qwen3 30B A3B 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 | Qwen3 30B A3B | Tencent Hy3 |
|---|---|---|
| Maker | Qwen | Tencent |
| Weights | Open weights | Open weights |
| Released | April 28, 2025 | July 6, 2026 |
| Context window | 41K | 262K |
| Input $/Mtok | $0.12 | $0.132 |
| Output $/Mtok | $0.50 | $0.528 |
| Coding index | Not benchmarked | Not benchmarked |
| Independent measurements behind it | 0 | 1 |
| Cost per solved task | Not computable | Not computable |
| Source | OpenRouter listing | OpenRouter listing |
Questions
Is Qwen3 30B A3B better than Hy3 for coding?
Neither model has a coding score yet, so the two cannot be ranked on coding. Pick Qwen3 30B A3B for output tokens 1.1x cheaper ($0.50 vs $0.528 per million) and work measured by LMArena Text: coding or LMArena Text: software and IT, where it leads.
Which is cheaper, Qwen3 30B A3B or Hy3?
Qwen3 30B A3B costs $0.12 per million input tokens and $0.50 per million output tokens; Hy3 costs $0.132 and $0.528. Cost per solved task is known for neither of them.
Which has the bigger context window, Qwen3 30B A3B or Hy3?
Hy3, with 262K tokens against 41K.
How many benchmarks is the Qwen3 30B A3B vs Hy3 comparison based on?
36 benchmarks have a result for both models as of October 22, 2026. Qwen3 30B A3B's coding index rests on 0 independent benchmarks and Hy3's on 1. Where only one model has a result, the page says so instead of naming a winner.