vs
3:0
overall capability wins (0 ties across 3 scored). Qwen3.8 Max 0902 leads overall.
- Pricing0:3
- Specs3:0
Verdict
Qwen3.8 Max 0902 offers ~6.3x larger context (1M vs 160K); Qwen3.8 Max 0902 allows ~8.0x more max output (131K vs 16K); Qwen3-Coder 30B-A3B ships open weights (Apache 2.0).
- Qwen3.8 Max 0902 offers ~6.3x larger context (1M vs 160K)
- Qwen3.8 Max 0902 allows ~8.0x more max output (131K vs 16K)
- Qwen3-Coder 30B-A3B ships open weights (Apache 2.0)
- Qwen3-Coder 30B-A3B is ~25.0x cheaper on a blended token basis than Qwen3.8 Max 0902
Cost for 1M in + 250K out
Illustrative chat workload at primary-provider list prices.
- Qwen3.8 Max 0902
- $3.50
- Qwen3-Coder 30B-A3B
- $0.14
Qwen3-Coder 30B-A3B is about 25.5x cheaper on this workload.
Price per 1M tokens
Full comparison
Organization
Alibaba
Alibaba
Family
Qwen
Qwen
License
Proprietary
Apache 2.0
Open weights
No
Yes
Release
Sep 2, 2026
Jul 22, 2025
Knowledge cutoff
-
-
API / provider
Alibaba Cloud
Alibaba / Together (ref.)
Modalities
text, image, video → text
text → text
Specs
Context window
1M
160K
Max output
131K
16K
Parameters
2.4T (95B act.)
30B (3B act.)
Pricing
Input $/1M
$2
$0.07
Output $/1M
$6
$0.27
Blended $/1M (3∶1)
$3
$0.12
Speed
tok/s
-
120
TTFT (s)
-
0.25
Reasoning
MMLU-Pro
—
70.6
GPQA Diamond
—
51.6
Humanity's Last Exam
—
3.8
AIME 2025
—
29.0
MATH-500
—
89.3
Humanity's Last Exam (with tools)
—
—
AA-Omniscience Accuracy
—
16.4
AA-LCR v1.1
—
32.7
CritPt
—
—
MMMU-Pro
—
—
IFBench
—
32.7
Chartography
—
—
Chartography (With Tools)
—
—
Coding
SWE-bench Verified
—
—
SWE-bench Pro
—
—
SWE-bench Multilingual
—
—
LiveCodeBench
—
40.3
Terminal-Bench 2.1
—
—
Aider Polyglot
—
—
Terminal-Bench 3
—
—
BigCodeBench
—
—
SciCode
—
27.8
CursorBench
—
—
SWE-Rebench
—
21.6
NL2Repo-Bench
—
—
DeepSWE
—
—
WebDev Arena
1,681
—
Terminal-Bench 4.0
—
—
CursorBench 4.0
—
—
FrontierCode v1.1 (Main)
—
—
Arena
LMArena Elo
—
—
| Metric | Qwen3.8 Max 0902 | Qwen3-Coder 30B-A3B | Delta |
|---|---|---|---|
| Identity | |||
| Organization | Alibaba | Alibaba | — |
| Family | Qwen | Qwen | — |
| License | Proprietary | Apache 2.0 | — |
| Open weights | No | Yes | — |
| Release | Sep 2, 2026 | Jul 22, 2025 | — |
| Knowledge cutoff | - | - | — |
| API / provider | Alibaba Cloud | Alibaba / Together (ref.) | — |
| Modalities | text, image, video → text | text → text | — |
| Specs | |||
| Context window | 1M | 160K | A +840K |
| Max output | 131K | 16K | A +115K |
| Parameters | 2.4T (95B act.) | 30B (3B act.) | A +2K |
| Pricing | |||
| Input $/1M | $2 | $0.07 | B +$1.93 |
| Output $/1M | $6 | $0.27 | B +$5.73 |
| Blended $/1M (3∶1) | $3 | $0.12 | B +$2.88 |
| Speed | |||
| tok/s | - | 120 | — |
| TTFT (s) | - | 0.25 | — |
| Reasoning | |||
| MMLU-Pro | — | 70.6 | — |
| GPQA Diamond | — | 51.6 | — |
| Humanity's Last Exam | — | 3.8 | — |
| AIME 2025 | — | 29.0 | — |
| MATH-500 | — | 89.3 | — |
| Humanity's Last Exam (with tools) | — | — | — |
| AA-Omniscience Accuracy | — | 16.4 | — |
| AA-LCR v1.1 | — | 32.7 | — |
| CritPt | — | — | — |
| MMMU-Pro | — | — | — |
| IFBench | — | 32.7 | — |
| Chartography | — | — | — |
| Chartography (With Tools) | — | — | — |
| Coding | |||
| SWE-bench Verified | — | — | — |
| SWE-bench Pro | — | — | — |
| SWE-bench Multilingual | — | — | — |
| LiveCodeBench | — | 40.3 | — |
| Terminal-Bench 2.1 | — | — | — |
| Aider Polyglot | — | — | — |
| Terminal-Bench 3 | — | — | — |
| BigCodeBench | — | — | — |
| SciCode | — | 27.8 | — |
| CursorBench | — | — | — |
| SWE-Rebench | — | 21.6 | — |
| NL2Repo-Bench | — | — | — |
| DeepSWE | — | — | — |
| WebDev Arena | 1,681 | — | — |
| Terminal-Bench 4.0 | — | — | — |
| CursorBench 4.0 | — | — | — |
| FrontierCode v1.1 (Main) | — | — | — |
| Arena | |||
| LMArena Elo | — | — | — |
Qwen3.8 Max 0902
- Context
- 1M / 131K out
- Parameters
- 2.4T (95B act.)
- Price
- $2 / $6
- Speed
- —
- Modalities
- text, image, video → text
- License
- Proprietary
September 2, 2026 hosted post-training snapshot of Qwen3.8-Max, further trained for coding and workplace/agent workflows. Same 2.4T MoE / 95B active architecture, 1M multimodal context, and $2/$6 QwenCloud pricing as the August 3 Max release. Arena Code/WebDev listed qwen3.8-max-0902 at 1688 (preliminary) on the Sep 2, 2026 snapshot.
Qwen3-Coder 30B-A3B
- Context
- 160K / 16K out
- Parameters
- 30B (3B act.)
- Price
- $0.07 / $0.27
- Speed
- 120 tok/s · 0.25s TTFT
- Modalities
- text → text
- License
- Apache 2.0
Sparse coding MoE in the Qwen3-Coder line — efficient agent edits without full 480B cost.
Benchmark charts
Winner bars are emphasized. Per-benchmark deltas sit above each chart.
These two models have no overlapping published benchmarks in our dataset. Compare specs and pricing instead.