Every AI leaderboard

28 leaderboards · 986 models ranked

Qwen3.8 2.4T A95B specs, prices, where it runs and its benchmark standing

open weights · #32 overall, podium score 2.9 · top 10 on 1 of 1 · released Aug 12, 2026

Facts

Provider
Alibaba
Release date
Aug 12, 2026
Boards
Parameters
2.45T total, 95B active per token
Hugging Face
Modalities
Text to Text
OpenRouter
License
Qwen3.8 Max
Hugging Face
Context window
1.05M tokens
Pricing
Pricing
$2 in / $6 out per 1M tokens
All prices
Hosting
4 priced API routes · OpenRouter · self-host: weights on Hugging Face
Pricing
Hardware
Estimate: about 4892 GB of weights as published (BF16), about 2446 GB at 8-bit, about 1223 GB at 4-bit, plus memory for the context
Hugging Face config
Weights
Open weights
Benchmarks
1 board captured, #32 overall
RankBoardScore
7MCP AtlasAgents and tools · as “Qwen3.8-2.4T-A95B (xHigh)”84.5%

Versions and release history

21 versions in this line since Sep 17, 2024, from Pricing, OpenRouter and the boards.

VersionReleased
Qwen3.8 27BAug 14, 20262 months ago
Qwen3.8 2.4T A95Bthis modelAug 12, 20262 months ago
Qwen3.6 27BApr 22, 20266 months ago
Qwen3.6 35B A3BApr 14, 20266 months ago
Qwen3.5 4BMar 2, 20268 months ago
Qwen3.5 27BFeb 25, 20268 months ago
Qwen3.5 122B A10BFeb 24, 20268 months ago
Qwen3.5 9BFeb 24, 20268 months ago
Qwen3.5-35B-A3BFeb 24, 20268 months ago
Qwen3.5 397B A17BFeb 13, 20268 months ago
Qwen3 30B A3BJul 30, 20251 year ago
Qwen3-235B-A22B-Thinking (Jul 2025)Jul 25, 20251 year ago
Category ranks, more models and catalog changes