Qwen3.8-Max
Canonical page on the main site: chinaaihub.com/models/qwen38-max
Provider
Model Family
Qwen3.8
Version
0902
Aliases
- qwen3.8-max-0902
- qwen3.8-max-2026-09-02
Release Date
2026-08
Status
active
Architecture
2.4T-parameter MoE, 95B activated, 512 experts (10 routed + 1 shared per token), 92 layers, Gated DeltaNet + Gated Attention hybrid
Parameter Information
- total_parameters: 2.4T
- active_parameters: 95B
Context Window
1048576
Maximum Output
131072
Capabilities
- reasoning: Yes
- coding: Yes
- vision: Yes
- video: Yes
- tool_calling: Yes
- structured_output: Yes
Open Weight
No
License
proprietary
Self Hosting
No
Api Available
Yes
Pricing
- input_price_per_1m: 2.0
- output_price_per_1m: 6.0
- currency: USD
- pricing_ref: alibaba-cloud
Official Api
Yes
Cloud Providers
- Alibaba Cloud
Regions
- china-beijing
- singapore
- hong-kong
- germany-frankfurt
- us-virginia
- japan-tokyo
Benchmark Results
| benchmark | score | metric | date | source_type | source_url |
|---|---|---|---|---|---|
| Terminal-Bench 2.1 | 86.6 | accuracy | 2026-08 | vendor_reported | https://huggingface.co/Qwen/Qwen3.8-2.4T-A95B |
| SWE-bench Pro | 67.7 | accuracy | 2026-08 | vendor_reported | https://huggingface.co/Qwen/Qwen3.8-2.4T-A95B |
| GPQA Diamond | 92.6 | accuracy | 2026-08 | vendor_reported | https://huggingface.co/Qwen/Qwen3.8-2.4T-A95B |
| HLE | 43.6 (56.2 with tools) | accuracy | 2026-08 | vendor_reported | https://huggingface.co/Qwen/Qwen3.8-2.4T-A95B |
| MRCR v2 256K | 92.9 | accuracy | 2026-08 | vendor_reported | https://huggingface.co/Qwen/Qwen3.8-2.4T-A95B |
Known Limitations
- Closed API model (the open Qwen3.8-2.4T-A95B weights are text-only and thinking-only - not the same product)
- Exact API release date not stated; the 0902 snapshot is dated 2026-09-02
- Prices differ by region: Singapore $2/$6; Beijing and Global regions $1.65/$4.951 per 1M tokens
- Benchmarks are from the vendor model card (Qwen3.8-Max column); not independently verified
Last Verified
2026-09-20
Sources
| source_name | source_url | source_type | last_verified | confidence |
|---|---|---|---|---|
| Model Studio — qwen3.8-max model detail | https://www.alibabacloud.com/help/en/model-studio/qwen3-8-max | official | 2026-09-20 | high |
| Model Studio — model pricing | https://www.alibabacloud.com/help/en/model-studio/model-pricing | official | 2026-09-20 | high |
| Hugging Face model card — Qwen3.8-2.4T-A95B | https://huggingface.co/Qwen/Qwen3.8-2.4T-A95B | official | 2026-09-20 | high |