Atlas

Models

← All models

Qwen3.5-Flash (source unspecified)

Alibaba

This benchmark entry preserves evaluations reported under Qwen3.5-Flash or a provider alias without an established dated checkpoint. Alibaba documents the current alias as equivalent to qwen3.5-flash-2026-02-23; evaluations that explicitly identify that snapshot are recorded separately. Exact release timing, context, and input modalities are left unspecified; reported reasoning controls remain on each run configuration.

Benchmark scores

BenchmarkScore
AIME 2024 (avg@8)93.3
AIME 2024-202592.5
AIME 2025 (avg@8)91.7
ALE-Bench826
Arena Code WebDev — Brand & Marketing1285
Arena Code WebDev — Consumer Product1225
Arena Code WebDev — Content Creation Tools1192
Arena Code WebDev — Data & Analytics1235
Arena Code WebDev — Frontend1243
Arena Code WebDev — Gaming1199
Arena Code WebDev — HTML1291
Arena Code WebDev — React1229
Arena Code WebDev — Reference-Based Design1241
Arena Code WebDev — Simulations1183
Arena Text — Business, Management & Financial Ops (No Style Control)1394
Arena Text — Chinese (No Style Control)1446
Arena Text — Creative Writing (No Style Control)1343
Arena Text — English (No Style Control)1404
Arena Text — Entertainment, Sports & Media (No Style Control)1347
Arena Text — Exclude Ties (No Style Control)1376
Arena Text — Exclude Ties (Style Controlled)1376
Arena Text — Expert (No Style Control)1407
Arena Text — French (No Style Control)1413
Arena Text — French (Style Controlled)1419
Arena Text — German (No Style Control)1391
Arena Text — Hard Prompts (No Style Control)1403
Arena Text — Hard Prompts English (No Style Control)1409
Arena Text — Instruction Following (No Style Control)1375
Arena Text — Japanese (No Style Control)1369
Arena Text — Korean (No Style Control)1344
Loading Atlas data…