Qwen: Qwen3.5-Flash
Released Feb 25, 2026Try it on:
The Qwen3.5 native vision-language Flash models are built on a hybrid architecture that integrates a linear attention mechanism with a sparse mixture-of-experts model, achieving higher inference efficiency. Compared to the...
Best overall rank
-
Across all subsets · overall
Total arena votes
0
Across 0 (subset, category) entries
Providers on OR
0
None yet
First seen
-
Awaiting first snapshot
Trend
No snapshot history yet - trends appear after the first daily refresh.
Pricing
Per token
- Input (per 1M)
- $0.065
- Output (per 1M)
- $0.260
Capabilities
Modality
text+image+video->textContext
1M tokens
Tokenizer
Qwen3
Max output
66K tokens
Moderation
Unmoderated
Supported parameters
include_reasoning
max_tokens
presence_penalty
reasoning
response_format
seed
structured_outputs
temperature
tool_choice
tools
top_p
Providers
No provider endpoint data yet.
Per-category ranks
No leaderboard entries for this model.