Qwen / MODEL EVIDENCE
48 benchmarks · 112 observations · 54 sources
48 benchmarks · large dot: Qwen3.8 Max; small dots: closest models where reported
Qwen 3.8 Max · GPQA Diamond · Accuracy ↗
92.73% · Qwen3.8 Max · GPQA Diamond · AA implementation · Artificial Analysis ↗
92.68% · Qwen3.8 Max (xhigh) · GPQA diamond · Best score (across scorers) · % · Epoch AI · GPQA diamond ↗
Qwen 3.8 Max · MMMU Pro · Accuracy ↗
82.31% · Qwen3.8 Max · MMMU-Pro · AA implementation · Artificial Analysis ↗
Qwen3.8 Max · τ³-Banking ↗
Qwen3.8 Max · Terminal-Bench 2.1 ↗
Qwen3.8 Max (xhigh) · FrontierMath-Tiers-1-3-v2-Private · Best score (across scorers) · % ↗
Qwen3.8 Max (xhigh) · FrontierMath-Tier-4-v2-Private · Best score (across scorers) · % ↗
Qwen3.8 Max (xhigh) · OTIS Mock AIME 2024-2025 · Best score (across scorers) · % ↗
Qwen3.8 Max (xhigh) · SimpleQA Verified · Best score (across scorers) · % ↗
Qwen3.8 Max (xhigh) · Chess Puzzles · Best score (across scorers) · % ↗
Qwen3.8 Max (xhigh) · Mystery Game Puzzles · Best score (across scorers) · % ↗
Qwen3.8 Max (xhigh) · DeepSWE · Pass@1 · % ↗
Qwen3.8 Max (xhigh) · LMCA · Score · % ↗
Qwen3.8 Max (xhigh) · DTBench · Accuracy · % ↗
Qwen3.8 Max (xhigh) · GDP.pdf · GDP.pdf score · % ↗
Qwen3.8 Max (xhigh) · FrontierSWE · Score · % ↗
Qwen 3.8 Max · General ECI · 26 October 2026 fit ↗
Qwen 3.8 Max (max) · Compression · reference-anchored score ↗
0.38 reference score · Qwen 3.8 Max (max) · LM training · reference-anchored score · Vals RSI Index · LM training ↗
0.14 reference score · Qwen 3.8 Max (max) · Harness engineering · reference-anchored score · Vals RSI Index · Harness engineering ↗
0.00 reference score · Qwen 3.8 Max (max) · Post-training · reference-anchored score · Vals RSI Index · Post-training ↗
Qwen 3.8 Max · Vals Multimodal Index · Accuracy ↗
Qwen 3.8 Max · Legal Research Bench · All-Pass ↗
Qwen 3.8 Max · LegalBench · Accuracy ↗
Qwen 3.8 Max · Harvey's Legal Agent Benchmark v1 · Accuracy ↗
Qwen 3.8 Max · Excel Modeling Benchmark · Accuracy ↗
60.07 % · Qwen 3.8 Max · Excel Modeling Benchmark v1 · Accuracy · Vals AI · Excel Modeling Benchmark v1 ↗
Qwen 3.8 Max · Tax Agent Bench · Accuracy ↗
Qwen 3.8 Max · Finance Agent v2 · Accuracy ↗
Qwen 3.8 Max · CorpFin · Accuracy ↗
Qwen 3.8 Max · MortgageTax · Accuracy ↗
75.55 % · Qwen 3.8 Max · TaxEval · Accuracy · Vals AI · TaxEval ↗
Qwen 3.8 Max · MedCode · Accuracy ↗
Qwen 3.8 Max · MedScribe · Accuracy ↗
Qwen 3.8 Max · Urgent safety · share of applicable ratings marked fail ↗
35% · Qwen 3.8 Max · Health advice · share of applicable ratings marked fail · Vals AI · SAFE-Teen ↗
51% · Qwen 3.8 Max · Relationship boundaries · share of applicable ratings marked fail · Vals AI · SAFE-Teen ↗
Qwen 3.8 Max · ProofBench · Accuracy ↗
Qwen 3.8 Max (Mini-SWE-agent) · Terminal-Bench-Science · Score ↗
Qwen 3.8 Max · MysteryMechanism · Accuracy ↗
Qwen 3.8 Max · MMLU Pro · Accuracy ↗
Qwen 3.8 Max · SAGE · Accuracy ↗
Qwen 3.8 Max · Vibe Code Bench · Vibe Code Bench 1–100 ↗
64.7 % · Qwen 3.8 Max (OpenHands) · Vibe Code Bench · original task set · Vals AI · Vibe Code Bench ↗
Qwen 3.8 Max · ProgramBench · Accuracy ↗
Qwen 3.8 Max (Mini-SWE-agent) · Terminal-Bench 4.0 · Score ↗
Qwen 3.8 Max (OpenHands) · SkillsBench · Score ↗
Qwen 3.8 Max · IOI · Accuracy ↗
Qwen 3.8 Max · SWE-bench Multimodal · Accuracy ↗
Qwen 3.8 Max · Code Migration v1 · Accuracy ↗
Qwen 3.8 Max (Mini-SWE-agent) · SWE-bench Verified · Score ↗
Qwen 3.8 Max · LiveCodeBench · Accuracy ↗
Qwen 3.8 Max · CyberBench · Accuracy ↗
Qwen 3.8 Max · Public Benefits Bench · Accuracy ↗
Qwen3.8-Max (max) · Humanity's Sixth Sense · Score ↗
qwen3.8-max · DistressBench · Score ↗
Qwen 3.8 Max · GDPx v1.0 · Overall · intended grading ↗
3.2% · Qwen 3.8 Max · GDPx v1.0 · Eight-domain subset · intended grading · GDPx v1.0 — SemiAnalysis ↗
9.0% · Qwen 3.8 Max · GDPx v1.0 · Recursive Self-Improvement · intended grading · GDPx v1.0 — SemiAnalysis ↗
0.1% · Qwen 3.8 Max · GDPx v1.0 · Chip Design · intended grading · GDPx v1.0 — SemiAnalysis ↗
7.4% · Qwen 3.8 Max · GDPx v1.0 · Cybersecurity · intended grading · GDPx v1.0 — SemiAnalysis ↗
5.1% · Qwen 3.8 Max · GDPx v1.0 · Biology · intended grading · GDPx v1.0 — SemiAnalysis ↗
0.0% · Qwen 3.8 Max · GDPx v1.0 · Finance · intended grading · GDPx v1.0 — SemiAnalysis ↗
0.0% · Qwen 3.8 Max · GDPx v1.0 · Legal · intended grading · GDPx v1.0 — SemiAnalysis ↗
11.2% · Qwen 3.8 Max · GDPx v1.0 · Quant Trading · intended grading · GDPx v1.0 — SemiAnalysis ↗
2.2% · Qwen 3.8 Max · GDPx v1.0 · Hardware Design · intended grading · GDPx v1.0 — SemiAnalysis ↗
3.0% · Qwen 3.8 Max · GDPx v1.0 · Materials Science · intended grading · GDPx v1.0 — SemiAnalysis ↗
0.0% · Qwen 3.8 Max · GDPx v1.0 · Alignment · intended grading · GDPx v1.0 — SemiAnalysis ↗
0.0% · Qwen 3.8 Max · GDPx v1.0 · Software Engineering · intended grading · GDPx v1.0 — SemiAnalysis ↗
9.2% · Qwen 3.8 Max · GDPx v1.0 · Overall · partial credit · GDPx v1.0 — SemiAnalysis ↗
9.9% · Qwen 3.8 Max · GDPx v1.0 · Eight-domain subset · partial credit · GDPx v1.0 — SemiAnalysis ↗
9.0% · Qwen 3.8 Max · GDPx v1.0 · Recursive Self-Improvement · partial credit · GDPx v1.0 — SemiAnalysis ↗
0.3% · Qwen 3.8 Max · GDPx v1.0 · Chip Design · partial credit · GDPx v1.0 — SemiAnalysis ↗
7.4% · Qwen 3.8 Max · GDPx v1.0 · Cybersecurity · partial credit · GDPx v1.0 — SemiAnalysis ↗
14.5% · Qwen 3.8 Max · GDPx v1.0 · Biology · partial credit · GDPx v1.0 — SemiAnalysis ↗
25.3% · Qwen 3.8 Max · GDPx v1.0 · Finance · partial credit · GDPx v1.0 — SemiAnalysis ↗
24.6% · Qwen 3.8 Max · GDPx v1.0 · Legal · partial credit · GDPx v1.0 — SemiAnalysis ↗
11.2% · Qwen 3.8 Max · GDPx v1.0 · Quant Trading · partial credit · GDPx v1.0 — SemiAnalysis ↗
2.2% · Qwen 3.8 Max · GDPx v1.0 · Hardware Design · partial credit · GDPx v1.0 — SemiAnalysis ↗
6.4% · Qwen 3.8 Max · GDPx v1.0 · Materials Science · partial credit · GDPx v1.0 — SemiAnalysis ↗
0.0% · Qwen 3.8 Max · GDPx v1.0 · Alignment · partial credit · GDPx v1.0 — SemiAnalysis ↗
0.0% · Qwen 3.8 Max · GDPx v1.0 · Software Engineering · partial credit · GDPx v1.0 — SemiAnalysis ↗
0.0% · Qwen 3.8 Max · GDPx v1.0 · Aria · Chip Design · intended grading · GDPx v1.0 — SemiAnalysis ↗
0.0% · Qwen 3.8 Max · GDPx v1.0 · Arcverex · Chip Design · intended grading · GDPx v1.0 — SemiAnalysis ↗
0.4% · Qwen 3.8 Max · GDPx v1.0 · Ergodex · Chip Design · intended grading · GDPx v1.0 — SemiAnalysis ↗
7.4% · Qwen 3.8 Max · GDPx v1.0 · Arithmetic · Cybersecurity · intended grading · GDPx v1.0 — SemiAnalysis ↗
5.1% · Qwen 3.8 Max · GDPx v1.0 · LatchBio · Biology · intended grading · GDPx v1.0 — SemiAnalysis ↗
0.0% · Qwen 3.8 Max · GDPx v1.0 · Living Context · Finance · intended grading · GDPx v1.0 — SemiAnalysis ↗
0.0% · Qwen 3.8 Max · GDPx v1.0 · Chakra · Legal · intended grading · GDPx v1.0 — SemiAnalysis ↗
11.2% · Qwen 3.8 Max · GDPx v1.0 · Third Moment · Quant Trading · intended grading · GDPx v1.0 — SemiAnalysis ↗
9.0% · Qwen 3.8 Max · GDPx v1.0 · Extrapolate · Recursive Self-Improvement · intended grading · GDPx v1.0 — SemiAnalysis ↗
2.2% · Qwen 3.8 Max · GDPx v1.0 · Beautiful Machines · Hardware Design · intended grading · GDPx v1.0 — SemiAnalysis ↗
0.0% · Qwen 3.8 Max · GDPx v1.0 · ValenceForge · Materials Science · intended grading · GDPx v1.0 — SemiAnalysis ↗
6.1% · Qwen 3.8 Max · GDPx v1.0 · Emulated · Materials Science · intended grading · GDPx v1.0 — SemiAnalysis ↗
0.0% · Qwen 3.8 Max · GDPx v1.0 · Fleet · Alignment · intended grading · GDPx v1.0 — SemiAnalysis ↗
0.0% · Qwen 3.8 Max · GDPx v1.0 · Datacurve · Software Engineering · intended grading · GDPx v1.0 — SemiAnalysis ↗
0.6% · Qwen 3.8 Max · GDPx v1.0 · Aria · Chip Design · partial credit · GDPx v1.0 — SemiAnalysis ↗
0.0% · Qwen 3.8 Max · GDPx v1.0 · Arcverex · Chip Design · partial credit · GDPx v1.0 — SemiAnalysis ↗
0.4% · Qwen 3.8 Max · GDPx v1.0 · Ergodex · Chip Design · partial credit · GDPx v1.0 — SemiAnalysis ↗
7.4% · Qwen 3.8 Max · GDPx v1.0 · Arithmetic · Cybersecurity · partial credit · GDPx v1.0 — SemiAnalysis ↗
14.5% · Qwen 3.8 Max · GDPx v1.0 · LatchBio · Biology · partial credit · GDPx v1.0 — SemiAnalysis ↗
25.3% · Qwen 3.8 Max · GDPx v1.0 · Living Context · Finance · partial credit · GDPx v1.0 — SemiAnalysis ↗
24.6% · Qwen 3.8 Max · GDPx v1.0 · Chakra · Legal · partial credit · GDPx v1.0 — SemiAnalysis ↗
11.2% · Qwen 3.8 Max · GDPx v1.0 · Third Moment · Quant Trading · partial credit · GDPx v1.0 — SemiAnalysis ↗
9.0% · Qwen 3.8 Max · GDPx v1.0 · Extrapolate · Recursive Self-Improvement · partial credit · GDPx v1.0 — SemiAnalysis ↗
2.2% · Qwen 3.8 Max · GDPx v1.0 · Beautiful Machines · Hardware Design · partial credit · GDPx v1.0 — SemiAnalysis ↗
4.0% · Qwen 3.8 Max · GDPx v1.0 · ValenceForge · Materials Science · partial credit · GDPx v1.0 — SemiAnalysis ↗
8.9% · Qwen 3.8 Max · GDPx v1.0 · Emulated · Materials Science · partial credit · GDPx v1.0 — SemiAnalysis ↗
0.0% · Qwen 3.8 Max · GDPx v1.0 · Fleet · Alignment · partial credit · GDPx v1.0 — SemiAnalysis ↗
0.0% · Qwen 3.8 Max · GDPx v1.0 · Datacurve · Software Engineering · partial credit · GDPx v1.0 — SemiAnalysis ↗
Similar positions on shared benchmark leaderboards. Smaller gaps are closer.
26 of 42 shared benchmark comparisons
30 of 42 shared benchmark comparisons
28 of 42 shared benchmark comparisons
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
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.
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.
Every plotted observation links to its evaluation. Reported configurations, benchmark versions and task subsets can differ; similarity describes shared scores, not training lift.
2026-08-02 · 2026-08-03
Qwen 3.8 Max · Qwen 3.8 Max (Mini-SWE-agent) · Qwen 3.8 Max (OpenHands) · Qwen 3.8 Max (max) · Qwen3.8 Max · Qwen3.8 Max (xhigh) · Qwen3.8-Max (max) · qwen3.8-max