GPT-5.4 Nano vs Qwen3.8 27B
Qwen3.8 27B leads GPT-5.4 Nano on coding, 74.2 to 44.8 on the Vibe Coding Index, and GPT-5.4 Nano costs 2.0x less per output token ($1.25 vs $2.55 per million). Updated October 22, 2026; 67 benchmarks have a result for both.
GPT-5.4 Nano or Qwen3.8 27B: which to pick
Pick Qwen3.8 27B for coding: it scores 74.2 on the Vibe Coding Index against 44.8 for GPT-5.4 Nano.
Pick GPT-5.4 Nano for output tokens 2.0x cheaper ($1.25 vs $2.55 per million) and a lower cost per solved task ($0.14 against $1.26).
- Shared results
- 67
- GPT-5.4 Nano only
- 18
- Qwen3.8 27B only
- 18
- Like-for-like categories
- 7 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
Qwen3.8 27B
Qwen3.8 27B scores 74.2 against 44.8.
2 and 4 benchmarks behind the scores
Intelligence
AA Intelligence Index
Qwen3.8 27B
Qwen3.8 27B leads, 33.7 against 20.7.
Same board, both measured
Agentic work
AA Agentic Index
Qwen3.8 27B
Qwen3.8 27B leads, 45.8 against 16.0.
Same board, both measured
Reasoning
GPQA Diamond
Qwen3.8 27B
Qwen3.8 27B leads, 88.9% against 77.5%.
Same board, both measured
Math
AIME
No call
Only GPT-5.4 Nano has a AIME result, 88.8%.
Directional only
Knowledge
MMLU Pro
Qwen3.8 27B
Qwen3.8 27B leads, 84.3% against 77.2%.
Same board, both measured
Long documents
Listed context window
Qwen3.8 27B
Qwen3.8 27B takes 1M tokens against 400K.
Listed by OpenRouter
Everyday cost
1K input + 500 output tokens
GPT-5.4 Nano
GPT-5.4 Nano $0.0008, Qwen3.8 27B $0.0017 per turn.
List prices
Cost per solved task
One SWE-bench Verified task, solved
GPT-5.4 Nano
GPT-5.4 Nano $0.14, Qwen3.8 27B $1.26.
Published attempt cost and pass rate
At a glance
- Coding index
- GPT-5.4 Nano 44.8 from 2 benchmarks, Qwen3.8 27B 74.2 from 4 benchmarks.
- Widest category
- Agentic work, 22.2 points apart on average over 5 shared benchmarks.
- Widest single gap
- Vibe Code Bench (Vals AI): GPT-5.4 Nano 26.1% against Qwen3.8 27B 64.8%, 38.7 points apart on a 0 to 100 scale.
- Output price
- GPT-5.4 Nano $1.25 against Qwen3.8 27B $2.55 per million output tokens, 2.0x apart.
- Context window
- GPT-5.4 Nano takes 400K tokens, Qwen3.8 27B 1M.
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 | GPT-5.4 Nano | Bars | Qwen3.8 27B | Ties |
|---|---|---|---|---|
| Coding | 3 | 6 | ||
| Intelligence | 0 | 2 | ||
| Agentic work | 0 | 5 | ||
| Reasoning | 0 | 11 | ||
| Math | 0 | 3 | ||
| Knowledge | 1 | 8 | ||
| General | 0 | 28 | ||
| All shared boards | 4 | 63 |
CodingLike-for-likeSix of these boards feed the Vibe Coding Index36
| Benchmark | GPT-5.4 Nano | Qwen3.8 27B | Leader |
|---|---|---|---|
| LiveBench coding | Not measured | 86.8% | Only Qwen3.8 27B measured |
| LMArena WebDev | Not measured | 1593 | Only Qwen3.8 27B measured |
| AA Coding Index | 56.1 | 68.1 | Qwen3.8 27B |
| LiveCodeBench (Vals AI) | 84.0% | 84.0% | GPT-5.4 Nano |
| SWE-bench Verified (Vals AI) | 69.8% | 86.0% | Qwen3.8 27B |
| Vibe Code Bench (Vals AI) | 26.1% | 64.8% | Qwen3.8 27B |
| Code migration (Vals AI) | 14.5% | 14.2% | GPT-5.4 Nano |
| ProgramBench (Vals AI) | Not measured | 0.0% | Only Qwen3.8 27B measured |
| IOI (Vals AI) | Not measured | 39.1% | Only Qwen3.8 27B measured |
| Terminal-Bench 2.1 (Vals AI) | 41.6% | 58.4% | Qwen3.8 27B |
| SkillsBench (Vals AI) | Not measured | 38.1% | Only Qwen3.8 27B measured |
| CyberBench (Vals AI) | Not measured | 67.8% | Only Qwen3.8 27B measured |
| SciCode | 47.2% | 46.6% | GPT-5.4 Nano |
| Terminal-Bench Hard (Artificial Analysis run) | 42.4% | Not measured | Only GPT-5.4 Nano measured |
| WeirdML | 49.2% | Not measured | Only GPT-5.4 Nano measured |
| ALE-Bench | 1004.5 | Not measured | Only GPT-5.4 Nano measured |
| LMArena Text: coding | 1404 | 1483 | Qwen3.8 27B |
| LMArena Text: software and IT | 1402 | 1478 | Qwen3.8 27B |
| LMArena WebDev: React | Not measured | 1602 | Only Qwen3.8 27B measured |
| LMArena WebDev: HTML | Not measured | 1570 | Only Qwen3.8 27B measured |
| LMArena WebDev: from an image | Not measured | 1577 | Only Qwen3.8 27B measured |
| Neither model has a result on the other 22 boards here | |||
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index02
| Benchmark | GPT-5.4 Nano | Qwen3.8 27B | Leader |
|---|---|---|---|
| AA Intelligence Index | 20.7 | 33.7 | Qwen3.8 27B |
| Epoch Capabilities Index | 146.0 | 151.7 | Qwen3.8 27B |
Agentic workLike-for-likePublic boards, shown as published, kept out of the index05
| Benchmark | GPT-5.4 Nano | Qwen3.8 27B | Leader |
|---|---|---|---|
| AA Agentic Index | 16.0 | 45.8 | Qwen3.8 27B |
| LiveBench research | Not measured | 59.4% | Only Qwen3.8 27B measured |
| LiveBench browser | Not measured | 32.0% | Only Qwen3.8 27B measured |
| LiveBench long agentic | Not measured | 27.9% | Only Qwen3.8 27B measured |
| Finance Agent (Vals AI) | 38.2% | 48.5% | Qwen3.8 27B |
| Tax Agent Bench (Vals AI) | 5.0% | 30.9% | Qwen3.8 27B |
| Legal Research Bench (Vals AI) | 6.3% | 36.1% | Qwen3.8 27B |
| Excel Modeling Benchmark (Vals AI) | 44.8% | 59.7% | Qwen3.8 27B |
| Terminal-Bench-Science (Vals AI) | Not measured | 0.0% | Only Qwen3.8 27B measured |
| τ²-Bench Telecom | 76.0% | Not measured | Only GPT-5.4 Nano measured |
| APEX-Agents | Not measured | 47.5% | Only Qwen3.8 27B measured |
| Neither model has a result on the other 31 boards here | |||
ReasoningLike-for-likePublic boards, shown as published, kept out of the index011
| Benchmark | GPT-5.4 Nano | Qwen3.8 27B | Leader |
|---|---|---|---|
| GPQA Diamond | 77.5% | 88.9% | Qwen3.8 27B |
| LiveBench reasoning | Not measured | 80.0% | Only Qwen3.8 27B measured |
| MMMU Pro (Vals AI) | 73.6% | 83.9% | Qwen3.8 27B |
| GPQA Diamond (Artificial Analysis run) | 81.7% | 90.5% | Qwen3.8 27B |
| Humanity's Last Exam (Artificial Analysis run) | 28.3% | 33.9% | Qwen3.8 27B |
| GPQA Diamond (Epoch AI run) | 78.5% | Not measured | Only GPT-5.4 Nano measured |
| ARC-AGI | 51.5% | 87.5% | Qwen3.8 27B |
| ARC-AGI-2 | 5.7% | 42.4% | Qwen3.8 27B |
| Chess puzzles | 45.0% | Not measured | Only GPT-5.4 Nano measured |
| Mystery game puzzles | 9.0% | Not measured | Only GPT-5.4 Nano measured |
| LMArena Text: hard prompts | 1381 | 1459 | Qwen3.8 27B |
| LMArena Text: expert | 1396 | 1479 | Qwen3.8 27B |
| LMArena Text: hard prompts in English | 1387 | 1461 | Qwen3.8 27B |
| LMArena Vision: diagrams | 1210 | 1285 | Qwen3.8 27B |
| LMArena Vision: homework | 1225 | 1295 | Qwen3.8 27B |
| Kagi LLM Benchmark | 39.7% | Not measured | Only GPT-5.4 Nano measured |
| Neither model has a result on the other 7 boards here | |||
MathLike-for-likePublic boards, shown as published, kept out of the index03
| Benchmark | GPT-5.4 Nano | Qwen3.8 27B | Leader |
|---|---|---|---|
| AIME | 88.8% | Not measured | Only GPT-5.4 Nano measured |
| OTIS Mock AIME | 87.8% | Not measured | Only GPT-5.4 Nano measured |
| FrontierMath, tiers 1 to 3 | 44.9% | Not measured | Only GPT-5.4 Nano measured |
| FrontierMath, tier 4 | 12.2% | Not measured | Only GPT-5.4 Nano measured |
| ProofBench | 5.0% | 16.0% | Qwen3.8 27B |
| LMArena Text: math | 1405 | 1455 | Qwen3.8 27B |
| LMArena Text: mathematical work | 1409 | 1458 | Qwen3.8 27B |
| Neither model has a result on the other 7 boards here | |||
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index18
| Benchmark | GPT-5.4 Nano | Qwen3.8 27B | Leader |
|---|---|---|---|
| MMLU Pro | 77.2% | 84.3% | Qwen3.8 27B |
| LegalBench (Vals AI) | 77.9% | 82.4% | Qwen3.8 27B |
| CaseLaw (Vals AI) | 51.9% | Not measured | Only GPT-5.4 Nano measured |
| CorpFin (Vals AI) | 61.2% | Not measured | Only GPT-5.4 Nano measured |
| TaxEval (Vals AI) | 67.4% | 70.8% | Qwen3.8 27B |
| MortgageTax (Vals AI) | 59.1% | 64.9% | Qwen3.8 27B |
| MedCode (Vals AI) | 41.0% | 28.7% | GPT-5.4 Nano |
| SimpleQA Verified | 11.7% | Not measured | Only GPT-5.4 Nano measured |
| LMArena Text: science | 1374 | 1467 | Qwen3.8 27B |
| LMArena Text: medicine and healthcare | 1369 | 1437 | Qwen3.8 27B |
| LMArena Text: legal and government | 1368 | 1447 | Qwen3.8 27B |
| LMArena Text: business and finance | 1366 | 1438 | Qwen3.8 27B |
| Neither model has a result on the other 8 boards here | |||
Web designDirectional onlyPublic boards, shown as published, kept out of the index1 board, none shared
| Benchmark | GPT-5.4 Nano | Qwen3.8 27B | Leader |
|---|---|---|---|
| LiveBench design | Not measured | 75.9% | Only Qwen3.8 27B measured |
| Neither model has a result on the other 8 boards here | |||
GeneralLike-for-likePublic boards, shown as published, kept out of the index028
| Benchmark | GPT-5.4 Nano | Qwen3.8 27B | Leader |
|---|---|---|---|
| LiveBench data analysis | Not measured | 76.6% | Only Qwen3.8 27B measured |
| SAGE (Vals AI) | 38.1% | 52.4% | Qwen3.8 27B |
| MedScribe (Vals AI) | 77.1% | 83.8% | Qwen3.8 27B |
| Vals Multimodal Index | 47.6% | Not measured | Only GPT-5.4 Nano measured |
| IFBench | 75.9% | Not measured | Only GPT-5.4 Nano measured |
| Long context reasoning (AA-LCR) | 76.7% | 82.0% | Qwen3.8 27B |
| DTBench | 80.3% | 88.0% | Qwen3.8 27B |
| LMCA | 36.9% | 41.4% | Qwen3.8 27B |
| Surface Evolver Bench | Not measured | 45.0% | Only Qwen3.8 27B measured |
| LMArena Text | 1372 | 1440 | Qwen3.8 27B |
| LMArena Text: instruction following | 1362 | 1436 | Qwen3.8 27B |
| LMArena Text: creative writing | 1315 | 1383 | Qwen3.8 27B |
| LMArena Vision | 1196 | 1269 | Qwen3.8 27B |
| LMArena Text: English | 1381 | 1448 | Qwen3.8 27B |
| LMArena Text: non-English | 1359 | 1428 | Qwen3.8 27B |
| LMArena Text: Chinese | 1391 | 1501 | Qwen3.8 27B |
| LMArena Text: German | 1368 | 1443 | Qwen3.8 27B |
| LMArena Text: Japanese | 1343 | 1387 | Qwen3.8 27B |
| LMArena Text: Korean | 1321 | 1384 | Qwen3.8 27B |
| LMArena Text: French | 1395 | 1465 | Qwen3.8 27B |
| LMArena Text: Spanish | 1371 | 1449 | Qwen3.8 27B |
| LMArena Text: Russian | 1363 | 1415 | Qwen3.8 27B |
| LMArena Text: Polish | 1366 | 1439 | Qwen3.8 27B |
| LMArena Text: multi-turn | 1382 | 1440 | Qwen3.8 27B |
| LMArena Text: longer queries | 1365 | 1448 | Qwen3.8 27B |
| LMArena Text: writing and language | 1338 | 1405 | Qwen3.8 27B |
| LMArena Text: entertainment and media | 1325 | 1393 | Qwen3.8 27B |
| LMArena Vision: English | 1202 | 1267 | Qwen3.8 27B |
| LMArena Vision: Chinese | 1232 | 1326 | Qwen3.8 27B |
| LMArena Vision: reading text | 1204 | 1281 | Qwen3.8 27B |
| LMArena Vision: humor | 1161 | 1235 | Qwen3.8 27B |
| LMArena Vision: recognizing things | 1141 | Not measured | Only GPT-5.4 Nano measured |
| LMArena Vision: creative writing | 1159 | 1237 | Qwen3.8 27B |
| Neither model has a result on the other 15 boards here | |||
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
| Benchmark | GPT-5.4 Nano | Qwen3.8 27B | 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
GPT-5.4 Nano costs less
Repository review
50K input + 3K output tokens
GPT-5.4 Nano costs less
Agent loop
200K input + 10K output tokens
GPT-5.4 Nano 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 | OpenAI GPT-5.4 Nano | Qwen3.8 27B |
|---|---|---|
| Maker | OpenAI | Qwen |
| Weights | No public weights listed | Open weights |
| Released | March 17, 2026 | August 14, 2026 |
| Context window | 400K | 1M |
| Input $/Mtok | $0.20 | $0.425 |
| Output $/Mtok | $1.25 | $2.55 |
| Coding index | 44.8 | 74.2 |
| Independent measurements behind it | 2 | 4 |
| Cost per solved task | $0.14 | $1.26 |
| Source | OpenRouter listing | OpenRouter listing |
Questions
Is GPT-5.4 Nano better than Qwen3.8 27B for coding?
Pick Qwen3.8 27B for coding: it scores 74.2 on the Vibe Coding Index against 44.8 for GPT-5.4 Nano. Pick GPT-5.4 Nano for output tokens 2.0x cheaper ($1.25 vs $2.55 per million) and a lower cost per solved task ($0.14 against $1.26).
Which is cheaper, GPT-5.4 Nano or Qwen3.8 27B?
GPT-5.4 Nano costs $0.20 per million input tokens and $1.25 per million output tokens; Qwen3.8 27B costs $0.425 and $2.55. Per solved SWE-bench Verified task, GPT-5.4 Nano costs $0.14 and Qwen3.8 27B $1.26.
Which has the bigger context window, GPT-5.4 Nano or Qwen3.8 27B?
Qwen3.8 27B, with 1M tokens against 400K.
How many benchmarks is the GPT-5.4 Nano vs Qwen3.8 27B comparison based on?
67 benchmarks have a result for both models as of October 22, 2026. GPT-5.4 Nano's coding index rests on 2 independent benchmarks and Qwen3.8 27B's on 4. Where only one model has a result, the page says so instead of naming a winner.