vs
6:6
overall capability wins (0 ties across 12 scored). Price/speed category wins do not decide overall. Tied overall.
- Reasoning4:2
- Coding0:2
- Arena1:0
- Tool use0:1
- Composite indices0:2
- Specs1:1
Verdict
Qwen3.8 27B leads coding (widest gap: +18.8 on TermBench); Qwen3.7 Plus edges reasoning & knowledge; Qwen3.7 Plus ranks higher on LMArena (+19 Elo).
- Qwen3.8 27B leads coding (widest gap: +18.8 on TermBench)
- Qwen3.7 Plus edges reasoning & knowledge
- Qwen3.7 Plus ranks higher on LMArena (+19 Elo)
- Qwen3.7 Plus offers ~3.8x larger context (1M vs 262K)
- Qwen3.8 27B allows ~2.0x more max output (131K vs 66K)
- Qwen3.8 27B ships open weights (Apache 2.0)
Cost for 1M in + 250K out
Illustrative chat workload at primary-provider list prices.
- Qwen3.7 Plus
- $0.80
- Qwen3.8 27B
- -
Price per 1M tokens
Full comparison
Organization
Alibaba
Alibaba
Family
Qwen
Qwen
License
Proprietary
Apache 2.0
Open weights
No
Yes
Release
Jun 3, 2026
Aug 14, 2026
Knowledge cutoff
-
-
API / provider
Alibaba Cloud
No primary API price listed
Modalities
text, image, video → text
text, image, video → text
Specs
Context window
1M
262K
Max output
66K
131K
Parameters
—
27B
Pricing
Input $/1M
$0.40
—
Output $/1M
$1.60
—
Blended $/1M (3∶1)
$0.70
-
Speed
tok/s
95
-
TTFT (s)
0.35
-
Reasoning
MMLU-Pro
—
—
GPQA Diamond
90.0
90.5
Humanity's Last Exam
35.6
33.9
AIME 2025
—
—
MATH-500
—
—
Humanity's Last Exam (with tools)
—
—
AA-Omniscience Accuracy
22.5
15.6
AA-LCR v1.1
73.0
82.0
CritPt
9.1
5.4
MMMU-Pro
80.5
76.3
IFBench
78.0
—
Chartography
—
—
Chartography (With Tools)
—
—
Coding
SWE-bench Verified
—
—
SWE-bench Pro
—
61.7
SWE-bench Multilingual
—
—
LiveCodeBench
—
90.3
Terminal-Bench 2.1
61.0
79.8
Aider Polyglot
—
—
Terminal-Bench 3
—
—
BigCodeBench
—
—
SciCode
46.1
46.6
CursorBench
—
—
SWE-Rebench
—
—
NL2Repo-Bench
—
42.3
DeepSWE
—
42.2
WebDev Arena
—
1,593
Terminal-Bench 4.0
—
—
CursorBench 4.0
—
—
FrontierCode v1.1 (Main)
—
—
Arena
LMArena Elo
1,456
1,437
| Metric | Qwen3.7 Plus | Qwen3.8 27B | Delta |
|---|---|---|---|
| Identity | |||
| Organization | Alibaba | Alibaba | — |
| Family | Qwen | Qwen | — |
| License | Proprietary | Apache 2.0 | — |
| Open weights | No | Yes | — |
| Release | Jun 3, 2026 | Aug 14, 2026 | — |
| Knowledge cutoff | - | - | — |
| API / provider | Alibaba Cloud | No primary API price listed | — |
| Modalities | text, image, video → text | text, image, video → text | — |
| Specs | |||
| Context window | 1M | 262K | A +738K |
| Max output | 66K | 131K | B +66K |
| Parameters | — | 27B | — |
| Pricing | |||
| Input $/1M | $0.40 | — | — |
| Output $/1M | $1.60 | — | — |
| Blended $/1M (3∶1) | $0.70 | - | — |
| Speed | |||
| tok/s | 95 | - | — |
| TTFT (s) | 0.35 | - | — |
| Reasoning | |||
| MMLU-Pro | — | — | — |
| GPQA Diamond | 90.0 | 90.5 | B +0.5 pts |
| Humanity's Last Exam | 35.6 | 33.9 | A +1.7 pts |
| AIME 2025 | — | — | — |
| MATH-500 | — | — | — |
| Humanity's Last Exam (with tools) | — | — | — |
| AA-Omniscience Accuracy | 22.5 | 15.6 | A +6.9 pts |
| AA-LCR v1.1 | 73.0 | 82.0 | B +9.0 pts |
| CritPt | 9.1 | 5.4 | A +3.7 pts |
| MMMU-Pro | 80.5 | 76.3 | A +4.2 pts |
| IFBench | 78.0 | — | — |
| Chartography | — | — | — |
| Chartography (With Tools) | — | — | — |
| Coding | |||
| SWE-bench Verified | — | — | — |
| SWE-bench Pro | — | 61.7 | — |
| SWE-bench Multilingual | — | — | — |
| LiveCodeBench | — | 90.3 | — |
| Terminal-Bench 2.1 | 61.0 | 79.8 | B +18.8 pts |
| Aider Polyglot | — | — | — |
| Terminal-Bench 3 | — | — | — |
| BigCodeBench | — | — | — |
| SciCode | 46.1 | 46.6 | B +0.5 pts |
| CursorBench | — | — | — |
| SWE-Rebench | — | — | — |
| NL2Repo-Bench | — | 42.3 | — |
| DeepSWE | — | 42.2 | — |
| WebDev Arena | — | 1,593 | — |
| Terminal-Bench 4.0 | — | — | — |
| CursorBench 4.0 | — | — | — |
| FrontierCode v1.1 (Main) | — | — | — |
| Arena | |||
| LMArena Elo | 1,456 | 1,437 | A +19 Elo |
Qwen3.7 Plus
- Context
- 1M / 66K out
- Parameters
- —
- Price
- $0.40 / $1.60
- Speed
- 95 tok/s · 0.35s TTFT
- Modalities
- text, image, video → text
- License
- Proprietary
Mid-tier Qwen3.7 multimodal agent (Jun 2026) — GUI grounding, tool use, and 1M context at roughly 5× lower cost than Max; no numeric scorecard retained without exact published values.
Qwen3.8 27B
- Context
- 262K / 131K out
- Parameters
- 27B
- Price
- —
- Speed
- —
- Modalities
- text, image, video → text
- License
- Apache 2.0
Open-weight Qwen3.8 27B dense vision-language model (Apache-2.0, released Aug 14, 2026) — native image and video understanding, 262K context extensible to 1M, flexible thinking control via reasoning_effort, and strong agentic coding for its size (Terminal-Bench 2.1 73.0, SWE-bench Pro 61.7, GPQA Diamond 89.2).
Benchmark charts
Winner bars are emphasized. Per-benchmark deltas sit above each chart.