vs
6:0
overall capability wins (1 ties across 7 scored). Qwen3.8 Flash Next leads overall.
- Reasoning4:0
- Coding1:0
- Pricing3:0
- Speed1:1
- Specs1:0
Verdict
Qwen3.8 Flash Next leads coding (widest gap: +12.3 on SciCode); Qwen3.8 Flash Next allows ~4.0x more max output (131K vs 33K); Qwen3.8 Flash Next ships open weights (Qwen Community License 1.0).
- Qwen3.8 Flash Next leads coding (widest gap: +12.3 on SciCode)
- Qwen3.8 Flash Next allows ~4.0x more max output (131K vs 33K)
- Qwen3.8 Flash Next ships open weights (Qwen Community License 1.0)
- Qwen3.8 Flash Next is ~7.1x cheaper on a blended token basis than Qwen3 Max
Cost for 1M in + 250K out
Illustrative chat workload at primary-provider list prices.
- Qwen3.8 Flash Next
- $0.27
- Qwen3 Max
- $1.75
Qwen3.8 Flash Next is about 6.5x cheaper on this workload.
Price per 1M tokens
Full comparison
Organization
Alibaba
Alibaba
Family
Qwen
Qwen
License
Qwen Community License 1.0
Proprietary
Open weights
Yes
No
Release
Aug 26, 2026
Sep 1, 2025
Knowledge cutoff
-
-
API / provider
Alibaba (QwenCloud)
Alibaba Cloud
Modalities
text, image, video → text
text, image → text
Specs
Context window
262K
262K
Max output
131K
33K
Parameters
—
—
Pricing
Input $/1M
$0.15
$0.50
Output $/1M
$0.47
$5
Blended $/1M (3∶1)
$0.23
$1.63
Speed
tok/s
77.3
65
TTFT (s)
2.79
0.5
Reasoning
MMLU-Pro
—
84.1
GPQA Diamond
92.3
76.4
Humanity's Last Exam
38.0
11.9
AIME 2025
—
80.7
MATH-500
—
—
Humanity's Last Exam (with tools)
—
—
AA-Omniscience Accuracy
24.5
24.4
AA-LCR v1.1
79.7
50.0
CritPt
11.1
—
MMMU-Pro
79.8
—
IFBench
—
44.1
Chartography
—
—
Chartography (With Tools)
—
—
Coding
SWE-bench Verified
—
—
SWE-bench Pro
—
—
SWE-bench Multilingual
—
—
LiveCodeBench
—
76.7
Terminal-Bench 2.1
86.1
—
Aider Polyglot
—
—
Terminal-Bench 3
—
—
BigCodeBench
—
—
SciCode
50.6
38.3
CursorBench
—
—
SWE-Rebench
—
—
NL2Repo-Bench
—
—
DeepSWE
—
—
WebDev Arena
1,635
—
Terminal-Bench 4.0
—
—
CursorBench 4.0
—
—
FrontierCode v1.1 (Main)
—
—
Arena
LMArena Elo
—
1,424
| Metric | Qwen3.8 Flash Next | Qwen3 Max | Delta |
|---|---|---|---|
| Identity | |||
| Organization | Alibaba | Alibaba | — |
| Family | Qwen | Qwen | — |
| License | Qwen Community License 1.0 | Proprietary | — |
| Open weights | Yes | No | — |
| Release | Aug 26, 2026 | Sep 1, 2025 | — |
| Knowledge cutoff | - | - | — |
| API / provider | Alibaba (QwenCloud) | Alibaba Cloud | — |
| Modalities | text, image, video → text | text, image → text | — |
| Specs | |||
| Context window | 262K | 262K | tie |
| Max output | 131K | 33K | A +98K |
| Parameters | — | — | — |
| Pricing | |||
| Input $/1M | $0.15 | $0.50 | A +$0.35 |
| Output $/1M | $0.47 | $5 | A +$4.53 |
| Blended $/1M (3∶1) | $0.23 | $1.63 | A +$1.40 |
| Speed | |||
| tok/s | 77.3 | 65 | A +12 |
| TTFT (s) | 2.79 | 0.5 | B +2.29 |
| Reasoning | |||
| MMLU-Pro | — | 84.1 | — |
| GPQA Diamond | 92.3 | 76.4 | A +15.9 pts |
| Humanity's Last Exam | 38.0 | 11.9 | A +26.1 pts |
| AIME 2025 | — | 80.7 | — |
| MATH-500 | — | — | — |
| Humanity's Last Exam (with tools) | — | — | — |
| AA-Omniscience Accuracy | 24.5 | 24.4 | A +0.1 pts |
| AA-LCR v1.1 | 79.7 | 50.0 | A +29.7 pts |
| CritPt | 11.1 | — | — |
| MMMU-Pro | 79.8 | — | — |
| IFBench | — | 44.1 | — |
| Chartography | — | — | — |
| Chartography (With Tools) | — | — | — |
| Coding | |||
| SWE-bench Verified | — | — | — |
| SWE-bench Pro | — | — | — |
| SWE-bench Multilingual | — | — | — |
| LiveCodeBench | — | 76.7 | — |
| Terminal-Bench 2.1 | 86.1 | — | — |
| Aider Polyglot | — | — | — |
| Terminal-Bench 3 | — | — | — |
| BigCodeBench | — | — | — |
| SciCode | 50.6 | 38.3 | A +12.3 pts |
| CursorBench | — | — | — |
| SWE-Rebench | — | — | — |
| NL2Repo-Bench | — | — | — |
| DeepSWE | — | — | — |
| WebDev Arena | 1,635 | — | — |
| Terminal-Bench 4.0 | — | — | — |
| CursorBench 4.0 | — | — | — |
| FrontierCode v1.1 (Main) | — | — | — |
| Arena | |||
| LMArena Elo | — | 1,424 | — |
Qwen3.8 Flash Next
- Context
- 262K / 131K out
- Parameters
- —
- Price
- $0.15 / $0.47
- Speed
- 77.3 tok/s · 2.79s TTFT
- Modalities
- text, image, video → text
- License
- Qwen Community License 1.0
Open-weight sparse-MoE Flash tier (6B active) previewing the Qwen4 architecture, released Aug 26, 2026 — 262K native context extensible to ~1M with YaRN, tuned for cost efficiency at $0.15/$0.47 per 1M (cache-hit $0.016). Reported total parameter count varies by source (~125-180B). Benchmark figures are Alibaba's self-reported launch numbers (SWE-bench Pro 62.5, HLE 35.9); Artificial Analysis places it at 56 on its Intelligence Index.
Qwen3 Max
- Context
- 262K / 33K out
- Parameters
- —
- Price
- $0.50 / $5
- Speed
- 65 tok/s · 0.5s TTFT
- Modalities
- text, image → text
- License
- Proprietary
Earlier Qwen Max API model; solid coding with competitive Chinese-market pricing.
Benchmark charts
Winner bars are emphasized. Per-benchmark deltas sit above each chart.