Qwen / qwen/qwen3.8-flash-free

Qwen3.8-Flash (Ucretsiz) - access through LLMTR

The same model as Qwen3.8-Flash, offered free for a limited period through a separate card. It provides an 800,000-token context and returns at most 131,072 tokens per response; it accepts image input alongside text, but not video. Tool calling, JSON object mode and schema-enforced output all work. Repeated prefixes are served from a prompt cache; since the card is free the gain is latency, not price. Reasoning is on by default and can be switched off per request; while it is on, the tool_choice forms that force a tool call ('required' or a named tool) are not supported, so turn reasoning off in the same request to use them. IMPORTANT: this card has a limit of only a few requests per minute that is SHARED across all users; when it is full you receive an error telling you when to retry. For sustained use, the metered qwen/qwen3.8-flash serves the same model without that limit and with a 1,000,000-token context. The free period ends on 22 September 2026.