backend new models check - alibaba - 2026-09-16

Files checked: src/backend/drivers/ai-chat/providers/alibaba/models.ts

chat (3)

• qwen3.8-flash | aliases: n/a | context: n/a | created: 2026-08-26 | price: see provider pricing page Notes: new Flash tier of the Qwen3.8 family — backend already has qwen3.8-max, qwen3.8-27b, qwen3.8-2.4t-a95b but not qwen3.8-flash. Genuinely missing, recent (3 weeks old).
• qwen-flash-character | aliases: n/a | context: n/a | created: 2026-04-10 | price: see provider pricing page Notes: character/roleplay-tuned deployment of qwen-flash; not present under any name in backend.
• qwen-plus-character | aliases: n/a | context: n/a | created: 2026-04-10 | price: see provider pricing page Notes: character/roleplay-tuned deployment of qwen-plus; not present under any name in backend.

Notes: qwen3.8-max-0902 (created 2026-09-02) is a dated snapshot of the already-backed qwen3.8-max — per the noise rule it is NOT new; suggest adding qwen3.8-max-0902 as an alias on the existing qwen3.8-max entry. Also excluded as non-chat modalities (image-gen, TTS, ASR, live-translate, omni-realtime voice, captioning, OCR, embeddings): 56 qwen-family entries (18 qwen-image-, 16 qwen3-tts-, 5 asr-/qwen-audio--asr-, 6 livetranslate-, 7 omni-*-realtime, plus qwen3-s2s-flash-realtime, qwen3-omni-30b-a3b-captioner, qwen-vl-ocr-2025-11-20, qwen3.7-text-embedding) — this backend is chat-only.

third-party rehosts (non-qwen), not real alibaba models: 24 legacy upstream models not in backend (verify intent): qwen-coder-plus (2025-10-14, pre-Qwen3 coder line, superseded by the already-backed qwen3-coder-plus/qwen3-coder-flash), qwen2-7b-instruct (2025-08-21, superseded Qwen2 generation)