Qwen3.8-27B
Alibaba · Sorti août 2026
Qwen3.8-27B est le modèle open-weight à 27 milliards de paramètres d'Alibaba, désormais disponible sous licence Apache 2.0 avec 262K de contexte natif et un déploiement sur une seule GPU. C'est le compagnon auto-hébergeable du flagship Qwen3.8-Max à 2,4T. Les benchmarks restent des chiffres d'Alibaba, d'où le verdict 'conditional' : un modèle réel et exécutable dont les gains de codage revendiqués par le fournisseur attendent encore une reproduction indépendante.
Est-ce fait pour vous ?
Recommandé pour
- Single-GPU self-hosting — 27B parameters fits a single consumer or datacenter GPU when quantized (4-bit: ~15GB VRAM, FP8: ~27GB, BF16: ~54GB)
- Open-weight ownership — full control over fine-tuning, data privacy, and deployment without API vendor lock-in
- Permissive commercial use: the weights are published on Hugging Face under Apache 2.0
- Long inputs on your own hardware: 262,144 tokens of native context, extensible up to 1,000,000 per the model card
- Cost-sensitive volume deployments — self-hosting breaks even vs. API at several billion tokens/month
Déconseillé pour
- Decisions that need independent benchmarks: the coding and agent gains are still Alibaba-reported
- Regulated environments requiring auditable model cards, verified safety evaluations, and independent evaluations
- Maximum capability workloads — the 27B model will not match the 2.4T Qwen3.8-Max flagship on complex reasoning or frontier benchmarks
- Teams who need predictable API pricing and SLAs from the model vendor: this entry covers the self-hosted open weights
Tarifs
Entrée
Free (open weights — self-hosted)
Sortie
Free (open weights — self-hosted)
Contexte
262K ctx; Apache 2.0; Alibaba API $0.5/$3 per 1M
Historique des verdicts
Sources
- Swfte AI: Qwen3.8-27B, The Version You Can Actually Run (Aug 4, 2026)août 2026
- DataNorth: Alibaba releases Qwen3.8-Max (Aug 4, 2026)août 2026
- MarkTechPost: Alibaba Qwen Releases Qwen3.8-Max (Aug 3, 2026)oct. 2026
- YottaLabs: Qwen 3.8 27B, Specs, Hardware Requirements, and How to Run It (Aug 4, 2026)août 2026
- Hugging Face Qwen3.8-27B model card (Apache 2.0) and Alibaba Cloud Model Studio pricingnov. 2026
- Digital Applied: Qwen3.8 Open Weights, Check This Before Downloading (Aug 7)août 2026
- Alibaba Cloud Model Studio: Model pricingnov. 2026
Journal de vérification
- Tarifs· Aucun changement
Agent automatisé
- Tarifs· Mis à jour
Agent automatisé
Alibaba Cloud Model Studio lists a hosted qwen3.8-27b at $0.5 input / $3 output per 1M (International, 1M free tokens for 90 days). Self-hosted weights remain free under Apache 2.0. pricing_context rewritten to '262K ctx; Apache 2.0; Alibaba API $0.5/$3 per 1M'; pricing_input/output labels untouched; sources gained the pricing page. Rating unchanged; no draft (additive hosted rate, not a price move).
- Contenu· Aucun changement
Agent automatisé