Qwen3-Coder 480B/A35B Instruct specs, prices, where it runs and its benchmark standing

Name: Qwen3-Coder 480B/A35B Instruct
Author: Alibaba

open weights · top 10 on 0 of 1 · released Jul 22, 2025

Provider: Alibaba
Release date: Jul 22, 2025; Boards
Parameters: 480B total, 35B active per token; Hugging Face
Modalities: Text to Text; OpenRouter
License: Apache 2.0; Hugging Face
Context window: 262K tokens; Pricing
Pricing: $0.3 in / $1 out per 1M tokens; All prices
Hosting: 2 priced API routes · AWS Bedrock (8 regions) · OpenRouter · self-host: weights on Hugging Face; Pricing Regions
Hardware: Estimate: about 960 GB of weights as published (BF16), about 480 GB at 8-bit, about 240 GB at 4-bit, plus memory for the context; Hugging Face config
Weights: Open weights
Benchmarks: 1 board captured

Rank	Board	Category	Score	Of	Over time
23	SWE-bench VerifiedCoding · stale since 2026-02-26 · as “Qwen3-Coder 480B/A35B Instruct”	Coding	55.4%	39

Versions and release history

4 versions in this line since Nov 12, 2024, from Pricing, OpenRouter and the boards.

Version	Released	Source
Qwen3 Coder 30B A3B Instruct	Jul 31, 20251 year ago	Pricing
Qwen3 Coder 480B A35B	Jul 23, 20251 year ago	OpenRouter
Qwen3-Coder 480B/A35B Instructthis model	Jul 22, 20251 year ago	Boards
Qwen2.5-Coder 32B Instruct	Nov 12, 20241 year ago	Boards

Category ranks, more models and catalog changes