Qwen3.8 2.4T A95B vs Qwen3 Next 80B A3B Thinking
Qwen3.8 2.4T A95B leads Qwen3 Next 80B A3B Thinking on the Artificial Analysis Intelligence Index, 39.9 to 11.2, and Qwen3 Next 80B A3B Thinking costs 5.0x less per output token ($1.20 vs $6.00 per million). Updated October 22, 2026; 2 benchmarks have a result for both.
Qwen3.8 2.4T A95B or Qwen3 Next 80B A3B Thinking: which to pick
Neither model has a coding score yet, so the two cannot be ranked on coding.
Pick Qwen3.8 2.4T A95B for a larger context window (1.05M against 262K tokens).
- Shared results
- 2
- Qwen3.8 2.4T A95B only
- 7
- Qwen3 Next 80B A3B Thinking only
- 28
- Like-for-like categories
- 2 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
Qwen3.8 2.4T A95B
Qwen3.8 2.4T A95B leads, 39.9 against 11.2.
Same board, both measured
Agentic work
AA Agentic Index
No call
Only Qwen3.8 2.4T A95B has a AA Agentic Index result, 50.1.
Directional only
Long documents
Listed context window
Qwen3.8 2.4T A95B
Qwen3.8 2.4T A95B takes 1.05M tokens against 262K.
Listed by OpenRouter
Everyday cost
1K input + 500 output tokens
Qwen3 Next 80B A3B Thinking
Qwen3.8 2.4T A95B $0.0050, Qwen3 Next 80B A3B Thinking $0.0008 per turn.
List prices
At a glance
- Coding index
- Qwen3.8 2.4T A95B not benchmarked, Qwen3 Next 80B A3B Thinking not benchmarked.
- Widest category
- Coding, 54.5 points apart on average over 1 shared benchmark.
- Widest single gap
- AA Coding Index: Qwen3.8 2.4T A95B 71.9 against Qwen3 Next 80B A3B Thinking 17.4, 54.5 points apart on a 0 to 100 scale.
- Output price
- Qwen3.8 2.4T A95B $6.00 against Qwen3 Next 80B A3B Thinking $1.20 per million output tokens, 5.0x apart.
- Context window
- Qwen3.8 2.4T A95B takes 1.05M tokens, Qwen3 Next 80B A3B Thinking 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 | Qwen3.8 2.4T A95B | Bars | Qwen3 Next 80B A3B Thinking | Ties |
|---|---|---|---|---|
| Coding | 1 | 0 | ||
| Intelligence | 1 | 0 | ||
| All shared boards | 2 | 0 |
CodingLike-for-likeSix of these boards feed the Vibe Coding Index10
| Benchmark | Qwen3.8 2.4T A95B | Qwen3 Next 80B A3B Thinking | Leader |
|---|---|---|---|
| AA Coding Index | 71.9 | 17.4 | Qwen3.8 2.4T A95B |
| SciCode | 54.1% | Not measured | Only Qwen3.8 2.4T A95B measured |
| LMArena Text: coding | Not measured | 1392 | Only Qwen3 Next 80B A3B Thinking measured |
| LMArena Text: software and IT | Not measured | 1393 | Only Qwen3 Next 80B A3B Thinking measured |
| Neither model has a result on the other 39 boards here | |||
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index10
| Benchmark | Qwen3.8 2.4T A95B | Qwen3 Next 80B A3B Thinking | Leader |
|---|---|---|---|
| AA Intelligence Index | 39.9 | 11.2 | Qwen3.8 2.4T A95B |
| Neither model has a result on the other 1 board here | |||
Agentic workDirectional onlyPublic boards, shown as published, kept out of the index2 boards, none shared
| Benchmark | Qwen3.8 2.4T A95B | Qwen3 Next 80B A3B Thinking | Leader |
|---|---|---|---|
| AA Agentic Index | 50.1 | Not measured | Only Qwen3.8 2.4T A95B measured |
| MCP Atlas | 84.5% | Not measured | Only Qwen3.8 2.4T A95B measured |
| Neither model has a result on the other 40 boards here | |||
ReasoningDirectional onlyPublic boards, shown as published, kept out of the index5 boards, none shared
| Benchmark | Qwen3.8 2.4T A95B | Qwen3 Next 80B A3B Thinking | Leader |
|---|---|---|---|
| GPQA Diamond (Artificial Analysis run) | 93.5% | Not measured | Only Qwen3.8 2.4T A95B measured |
| Humanity's Last Exam (Artificial Analysis run) | 42.4% | Not measured | Only Qwen3.8 2.4T A95B measured |
| LMArena Text: hard prompts | Not measured | 1371 | Only Qwen3 Next 80B A3B Thinking measured |
| LMArena Text: expert | Not measured | 1376 | Only Qwen3 Next 80B A3B Thinking measured |
| LMArena Text: hard prompts in English | Not measured | 1384 | Only Qwen3 Next 80B A3B Thinking measured |
| Neither model has a result on the other 18 boards here | |||
MathDirectional onlyPublic boards, shown as published, kept out of the index3 boards, none shared
| Benchmark | Qwen3.8 2.4T A95B | Qwen3 Next 80B A3B Thinking | Leader |
|---|---|---|---|
| LMArena Text: math | Not measured | 1399 | Only Qwen3 Next 80B A3B Thinking measured |
| LMArena Text: mathematical work | Not measured | 1390 | Only Qwen3 Next 80B A3B Thinking measured |
| MathArena | 31.6% | Not measured | Only Qwen3.8 2.4T A95B measured |
| Neither model has a result on the other 11 boards here | |||
KnowledgeDirectional onlyPublic boards, shown as published, kept out of the index4 boards, none shared
| Benchmark | Qwen3.8 2.4T A95B | Qwen3 Next 80B A3B Thinking | Leader |
|---|---|---|---|
| LMArena Text: science | Not measured | 1384 | Only Qwen3 Next 80B A3B Thinking measured |
| LMArena Text: medicine and healthcare | Not measured | 1392 | Only Qwen3 Next 80B A3B Thinking measured |
| LMArena Text: legal and government | Not measured | 1364 | Only Qwen3 Next 80B A3B Thinking measured |
| LMArena Text: business and finance | Not measured | 1360 | Only Qwen3 Next 80B A3B Thinking measured |
| Neither model has a result on the other 16 boards here | |||
Web designNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
| Benchmark | Qwen3.8 2.4T A95B | Qwen3 Next 80B A3B Thinking | Leader |
|---|---|---|---|
| Neither model has a result on 9 boards here | |||
GeneralDirectional onlyPublic boards, shown as published, kept out of the index18 boards, none shared
| Benchmark | Qwen3.8 2.4T A95B | Qwen3 Next 80B A3B Thinking | Leader |
|---|---|---|---|
| Long context reasoning (AA-LCR) | 80.3% | Not measured | Only Qwen3.8 2.4T A95B measured |
| LMArena Text | Not measured | 1368 | Only Qwen3 Next 80B A3B Thinking measured |
| LMArena Text: instruction following | Not measured | 1344 | Only Qwen3 Next 80B A3B Thinking measured |
| LMArena Text: creative writing | Not measured | 1314 | Only Qwen3 Next 80B A3B Thinking measured |
| LMArena Text: English | Not measured | 1392 | Only Qwen3 Next 80B A3B Thinking measured |
| LMArena Text: non-English | Not measured | 1342 | Only Qwen3 Next 80B A3B Thinking measured |
| LMArena Text: Chinese | Not measured | 1407 | Only Qwen3 Next 80B A3B Thinking measured |
| LMArena Text: German | Not measured | 1356 | Only Qwen3 Next 80B A3B Thinking measured |
| LMArena Text: Japanese | Not measured | 1280 | Only Qwen3 Next 80B A3B Thinking measured |
| LMArena Text: Korean | Not measured | 1310 | Only Qwen3 Next 80B A3B Thinking measured |
| LMArena Text: French | Not measured | 1351 | Only Qwen3 Next 80B A3B Thinking measured |
| LMArena Text: Spanish | Not measured | 1358 | Only Qwen3 Next 80B A3B Thinking measured |
| LMArena Text: Russian | Not measured | 1337 | Only Qwen3 Next 80B A3B Thinking measured |
| LMArena Text: Polish | Not measured | 1345 | Only Qwen3 Next 80B A3B Thinking measured |
| LMArena Text: multi-turn | Not measured | 1346 | Only Qwen3 Next 80B A3B Thinking measured |
| LMArena Text: longer queries | Not measured | 1353 | Only Qwen3 Next 80B A3B Thinking measured |
| LMArena Text: writing and language | Not measured | 1330 | Only Qwen3 Next 80B A3B Thinking measured |
| LMArena Text: entertainment and media | Not measured | 1312 | Only Qwen3 Next 80B A3B Thinking measured |
| Neither model has a result on the other 30 boards here | |||
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
| Benchmark | Qwen3.8 2.4T A95B | Qwen3 Next 80B A3B 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
Qwen3 Next 80B A3B Thinking costs less
Repository review
50K input + 3K output tokens
Qwen3 Next 80B A3B Thinking costs less
Agent loop
200K input + 10K output tokens
Qwen3 Next 80B A3B 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 | Qwen3.8 2.4T A95B | Qwen3 Next 80B A3B Thinking |
|---|---|---|
| Maker | Qwen | Qwen |
| Weights | Open weights | Open weights |
| Released | August 12, 2026 | September 11, 2025 |
| Context window | 1.05M | 262K |
| Input $/Mtok | $2.00 | $0.15 |
| Output $/Mtok | $6.00 | $1.20 |
| Coding index | Not benchmarked | Not benchmarked |
| Independent measurements behind it | 0 | 0 |
| Cost per solved task | Not computable | Not computable |
| Source | OpenRouter listing | OpenRouter listing |
Questions
Is Qwen3.8 2.4T A95B better than Qwen3 Next 80B A3B Thinking for coding?
Neither model has a coding score yet, so the two cannot be ranked on coding. Pick Qwen3.8 2.4T A95B for a larger context window (1.05M against 262K tokens).
Which is cheaper, Qwen3.8 2.4T A95B or Qwen3 Next 80B A3B Thinking?
Qwen3.8 2.4T A95B costs $2.00 per million input tokens and $6.00 per million output tokens; Qwen3 Next 80B A3B Thinking costs $0.15 and $1.20. Cost per solved task is known for neither of them.
Which has the bigger context window, Qwen3.8 2.4T A95B or Qwen3 Next 80B A3B Thinking?
Qwen3.8 2.4T A95B, with 1.05M tokens against 262K.
How many benchmarks is the Qwen3.8 2.4T A95B vs Qwen3 Next 80B A3B Thinking comparison based on?
2 benchmarks have a result for both models as of October 22, 2026. Qwen3.8 2.4T A95B's coding index rests on 0 independent benchmarks and Qwen3 Next 80B A3B Thinking's on 0. Where only one model has a result, the page says so instead of naming a winner.