Hybrid Gated DeltaNet/attention ultra-sparse MoE (80B total, 3B active) in Instruct and Thinking variants.
Released
Sep 11, 2025
Context window
262K tokens
Max output
Not disclosed
Parameters
80B 3B active
Input price
Not disclosed
Output price
Not disclosed
Prices are first-party list prices in USD at the standard tier. Last verified Oct 30, 2026.
Pricing
Launch priceSep 11, 2025
$0.15 in / $1.20 out
Current price
No longer sold
Specifications
- Developer
- Alibaba
- Model type
- Text LLM
- Architecture
- Hybrid
- Parameters
- 80B (3B active)
- Input
- Text
- Output
- Text
- Knowledge cutoff
- Not disclosed
- License
- Apache License 2.0
- API access
- No first-party API
- Capabilities
- Long context Reasoning