Alibaba Qwen Open-Sources Flagship Qwen3.8-2.4T-A95B with Trillion Parameters
智东西 · wechat · 2026-08-13
Alibaba's Qwen team open-sourced the weights for its flagship model, Qwen3.8-2.4T-A95B, featuring 2.4 trillion total parameters with 95 billion activated per token. It natively supports over 1 million tokens of context.
Capabilities & Benchmarks
The model excels in coding and long-horizon agentic tasks, topping the PaperBench and OSWorld benchmarks. However, it trails slightly behind GPT-5.6Sol and Fable5 on TerminalBench2.1 and SWE-benchPro. Qwen demonstrated the model's autonomous capabilities by having it code continuously for 16 days and manage a virtual e-commerce business.
Deployment & Pricing
The model defaults to a mandatory thinking mode. UnslothAI managed to compress the model from 4.9TB to 397GB using dynamic quantization, significantly lowering local deployment barriers. Overseas API pricing is set at $2 per million input tokens and $6 per million output tokens.
More from Models
- Unverified claim: Apple and Meta's 816B-parameter Muse Spark Open runs on two Mac Studios — TansuYegen · 2026-10-16
- Gemini 4 model card: model eliminated months of trial-and-error on bioweapons research — Puzzleheaded-King584 · 2026-10-16
- Every AI picks 7: Anthropic engineer's rule for escaping model taste convergence — rubenhassid · 2026-10-16
- Brain teaser: strip all attention, MLP and embedding matrices from a Llama 3 — how many trainable params remain? — Sauers_ · 2026-10-16
- Lean v4.35.0 ships verified kernels — one written and proven consistent by Claude — geoffreyirving · 2026-10-16
- One flagged seat froze an entire Claude team org — founders share backup plans — RexDouglass · 2026-10-16