Qwen3.8 Max (0902) · Intelligence Index v5.0 ↗
1 more reported result
45.42 index · Qwen3.8 Max (0902) · Intelligence Index v4.3.2 · Artificial Analysis ↗
Qwen / MODEL EVIDENCE
Qwen3.8 Max (0902)
47 benchmarks · 55 observations · 49 sources
Scores at a glance
47 benchmarks · large dot: Qwen3.8 Max (0902); small dots: closest models where reported
AA-Briefcase v1.1 ↗1607.70 Elo
Qwen3.8 Max (0902) · AA-Briefcase v1.1 · overall Elo ↗
AutomationBench-AA ↗56.19%
Qwen3.8 Max (0902) · AutomationBench-AA · guardrail-gated objective score ↗
GDP.pdf ↗22.80%
Qwen3.8 Max (0902) · GDP.pdf · all-pass ↗
Terminal-Bench 4.0 ↗38.89%
Qwen3.8 Max (0902) · Terminal-Bench 4.0 · AA harness ↗
1 more reported result
38.89 % · Qwen 3.8 Max 0902 (Mini-SWE-agent) · Terminal-Bench 4.0 · Score · Vals AI · Terminal-Bench 4.0 ↗
Qwen3.8 Max (0902) · Terminal-Bench-Science 0.1 · AA harness ↗
FrontierCode 2.0 ↗23.42%
Qwen3.8 Max (0902) · FrontierCode 2.0 · AA implementation ↗
Qwen3.8 Max (0902) · CritPt v2 · hard subset · AA implementation ↗
HLE-Diamond ↗24.00%
Qwen3.8 Max (0902) · HLE-Diamond · AA implementation ↗
AA-Omniscience v1.1 ↗10.47 index
Qwen3.8 Max (0902) · AA-Omniscience v1.1 · error-penalized index ↗
1 more reported result
11.98 index · Qwen3.8 Max (0902) · AA-Omniscience · error-penalized score · Artificial Analysis ↗
Qwen3.8 Max (0902) · AA-Omniscience v1.1 · accuracy ↗
1 more reported result
31.68% · Qwen3.8 Max (0902) · AA-Omniscience · accuracy · Artificial Analysis ↗
Qwen3.8 Max (0902) · AA-Omniscience v1.1 · hallucination rate, lower is better ↗
1 more reported result
28.84% · Qwen3.8 Max (0902) · AA-Omniscience · hallucination rate, lower is better · Artificial Analysis ↗
GDPval-AA v2.1 ↗1666.26 Elo
Qwen3.8 Max (0902) · GDPval-AA v2.1 · Elo ↗
AA-3D ↗944.77 Elo
Qwen3.8 Max (0902) · AA-3D · Elo ↗
APEX-Agents-AA ↗42.40%
Qwen3.8 Max (0902) · APEX-Agents-AA ↗
AA-AnalystAgent ↗45.00%
Qwen3.8 Max (0902) · AA-AnalystAgent · strict score ↗
EnterpriseOps-Gym-AA ↗47.63%
Qwen3.8 Max (0902) · EnterpriseOps-Gym-AA ↗
FrontierCode 1.1 ↗43.16%
Qwen3.8 Max (0902) · FrontierCode 1.1 · AA implementation ↗
AA-LCR v1.1 ↗80.33%
Qwen3.8 Max (0902) · AA-LCR v1.1 ↗
SciCode ↗52.08%
Qwen3.8 Max (0902) · SciCode · AA implementation ↗
Humanity’s Last Exam ↗43.10%
Qwen3.8 Max (0902) · Humanity’s Last Exam · AA implementation ↗
GPQA Diamond ↗92.83%
Qwen3.8 Max (0902) · GPQA Diamond · AA implementation ↗
1 more reported result
92.3% · Qwen3.8 Max (0902) (xhigh) · GPQA diamond · Best score (across scorers) · % · Epoch AI · GPQA diamond ↗
ITBench-AA ↗40.25%
Qwen3.8 Max (0902) · ITBench-AA · root-cause score ↗
MMMU-Pro ↗82.77%
Qwen3.8 Max (0902) · MMMU-Pro · AA implementation ↗
MLCR-AA ↗20.00%
Qwen3.8 Max (0902) · MLCR-AA · overall ↗
τ³-Banking ↗47.84%
Qwen3.8 Max (0902) · τ³-Banking ↗
Terminal-Bench 2.1 ↗88.76%
Qwen3.8 Max (0902) · Terminal-Bench 2.1 ↗
CritPt ↗17.71%
Qwen3.8 Max (0902) · CritPt · original ↗
Qwen3.8 Max (0902) (xhigh) · FrontierMath-Tiers-1-3-v2-Private · Best score (across scorers) · % ↗
Qwen3.8 Max (0902) (xhigh) · FrontierMath-Tier-4-v2-Private · Best score (across scorers) · % ↗
Qwen3.8 Max (0902) (xhigh) · OTIS Mock AIME 2024-2025 · Best score (across scorers) · % ↗
SimpleQA Verified ↗47.3%
Qwen3.8 Max (0902) (xhigh) · SimpleQA Verified · Best score (across scorers) · % ↗
Qwen3.8 Max (0902) (xhigh) · Chess Puzzles · Best score (across scorers) · % ↗
Qwen3.8 Max (0902) (xhigh) · Furniture Assembly · Best score (across scorers) · % ↗
FrontierSWE v2 ↗17.81%
Qwen3.8 Max (0902) (xhigh) · FrontierSWE · Score · % ↗
Epoch Capabilities Index (ECI) ↗155.08 ECI
Qwen3.8 Max (0902) · General ECI · 26 October 2026 fit ↗
Vals Index ↗59.05 %
Qwen 3.8 Max 0902 · Vals Index · Accuracy ↗
Vals Finance Index ↗61.66 %
Qwen 3.8 Max 0902 · Vals Finance Index · Accuracy ↗
Vals Legal Index ↗73.9 %
Qwen 3.8 Max 0902 · Vals Legal Index · Accuracy ↗
Legal Research Bench ↗46.15 %
Qwen 3.8 Max 0902 · Legal Research Bench · All-Pass ↗
HLAB-Vals-1.1 ↗15.42 %
Qwen 3.8 Max 0902 · HLAB-Vals-1.1 · Task Pass Rate ↗
Vals Finance Agent v2.1 ↗58.83 %
Qwen 3.8 Max 0902 · Finance Agent · Accuracy ↗
Qwen 3.8 Max 0902 · Excel Modeling Benchmark · Accuracy ↗
1 more reported result
71.98 % · Qwen 3.8 Max 0902 · Excel Modeling Benchmark v1 · Accuracy · Vals AI · Excel Modeling Benchmark v1 ↗
Tax Agent Bench ↗49.66 %
Qwen 3.8 Max 0902 · Tax Agent Bench · Accuracy ↗
Vals Finance Agent v2 ↗57.87 %
Qwen 3.8 Max 0902 · Finance Agent v2 · Accuracy ↗
Vibe Code Bench ↗24.83 %
Qwen 3.8 Max 0902 · Vibe Code Bench · Vibe Code Bench 1–100 ↗
1 more reported result
77.14 % · Qwen 3.8 Max 0902 (OpenHands) · Vibe Code Bench · original task set · Vals AI · Vibe Code Bench ↗
Code Migration v1.1 ↗57.19 %
Qwen 3.8 Max 0902 · Code Migration · Accuracy ↗
Closest models by score
Similar positions on shared benchmark leaderboards. Smaller gaps are closer.
DeepSeek V4.1 Flash 1017 →DeepSeek
34 of 44 shared benchmark comparisons
Grok 4.7 →xAI
37 of 44 shared benchmark comparisons
GLM-5.3 →Z.ai
30 of 44 shared benchmark comparisons
How similarity is measured
Mean absolute gap in percentile positions on shared benchmark settings; lower is closer. At least 60% overlap is required, with a minimum of three when available. Independent evaluations are used when there are at least three; otherwise published report comparisons are used. Repeated model observations within one setting are averaged before ranking. Writing-style traits and composite indices are excluded. This describes similarity, not an overall quality ranking.
POST-TRAINING / SOURCE DISCLOSURE
What the Qwen3.8 Max / 2.4T-A95B report says
The official model card identifies 2.4T-A95B as post-trained and distinguishes its open weights from the Max service, which adds features. It documents the release and architecture without a reproducible optimization recipe.
- Post-trained open-weight checkpoint
- Separate Max service configuration
Supervised fine-tuning · background →Preference learning · background →Stateful agent training · background →
No named optimizer, reward mixture or stage-by-stage training schedule is specified in this card. The method links are background, not claims of adoption; open weights and the Max service remain separate model identities.
Sources
Every plotted observation links to its evaluation. Reported configurations, benchmark versions and task subsets can differ; similarity describes shared scores, not training lift.
- AA-3D: Interactive 3D Experience Benchmark | Artificial Analysis ↗
- AA-AnalystAgent Benchmark | Artificial Analysis ↗
- AA-Briefcase v1.1: Agentic Knowledge Work Benchmark | Artificial Analysis ↗
- APEX-Agents-AA Benchmark | Artificial Analysis ↗
- Artificial Analysis Intelligence Index v5.0 — Artificial Analysis ↗
- AA-LCR v1.1: Long Context Reasoning Benchmark | Artificial Analysis ↗
- AutomationBench-AA: Agentic SaaS Workflow Benchmark | Artificial Analysis ↗
- CritPt Benchmark | Artificial Analysis ↗
- CritPt v2 — Hard Subset — Artificial Analysis ↗
- EnterpriseOps-Gym-AA Benchmark | Artificial Analysis ↗
- FrontierCode 1.1 Benchmark | Artificial Analysis ↗
- FrontierCode 2.0 Benchmark — Artificial Analysis ↗
- GDP.pdf Benchmark | Artificial Analysis ↗
- GDPval-AA v2.1 Benchmark | Artificial Analysis ↗
- GPQA Diamond Benchmark | Artificial Analysis ↗
- HLE-Diamond Benchmark — Artificial Analysis ↗
- Humanity's Last Exam Benchmark | Artificial Analysis ↗
- ITBench-AA Benchmark | Artificial Analysis ↗
- MLCR-AA: Medical Long Context Reasoning Benchmark | Artificial Analysis ↗
- MMMU-Pro Benchmark | Artificial Analysis ↗
- AA-Omniscience v1.1: Knowledge and Hallucination Benchmark — Artificial Analysis ↗
- SciCode Benchmark | Artificial Analysis ↗
- 𝜏³-Banking Benchmark | Artificial Analysis ↗
- Terminal-Bench-Science 0.1 Benchmark | Artificial Analysis ↗
- Terminal-Bench 2.1 Benchmark | Artificial Analysis ↗
- Terminal-Bench 4.0 Benchmark | Artificial Analysis ↗
- Chess Puzzles ↗
- FrontierMath Tier 4 (v2) ↗
- FrontierMath Tiers 1-3 (v2) ↗
- Furniture Assembly Benchmark ↗
- GPQA Diamond ↗
- OTIS Mock AIME 2024-2025 ↗
- SimpleQA Verified ↗
- AI benchmarks & capabilities ↗
- FrontierSWE — software engineering benchmark ↗
- Code Migration — evaluation methodology and results ↗
- Excel Modeling Benchmark v1 — evaluation methodology and results ↗
- Excel Modeling Benchmark — evaluation methodology and results ↗
- Finance Agent v2 — evaluation methodology and results ↗
- Finance Agent — evaluation methodology and results ↗
- Vals Finance Index — evaluation methodology and results ↗
- HLAB-Vals-1.1 — evaluation methodology and results ↗
- Vals Legal Index — evaluation methodology and results ↗
- Legal Research Bench — evaluation methodology and results ↗
- Tax Agent Bench — evaluation methodology and results ↗
- Terminal-Bench 4.0 — evaluation methodology and results ↗
- Vals Index — evaluation methodology and results ↗
- Vibe Code Bench 1-100 — evaluation methodology and results ↗
- Vibe Code Bench — evaluation methodology and results ↗
Release dates reported by sources
2026-09-01 · 2026-09-02
Names used in source tables
Qwen 3.8 Max 0902 · Qwen 3.8 Max 0902 (Mini-SWE-agent) · Qwen 3.8 Max 0902 (OpenHands) · Qwen3.8 Max (0902) · Qwen3.8 Max (0902) (xhigh)