Qwen3 Coder 480B A35B
Qwen3 Coder 480B A35B is Alibaba's largest coding-specialized MoE model with 480B total and 35B active parameters, native 256K token context extended to 1M.
Model Overview
Capabilities, design details, and architectural traits
Qwen3 Coder 480B A35B - Qwen's Largest Agentic Coding Model
Qwen3 Coder 480B A35B-Instruct is an agentic code model, designed for autonomous software engineering, repository-level understanding, and multi-turn tool use with interactive reasoning.
What Makes It Different: Agentic Coding with Massive Context Window
Qwen3 Coder 480B A35B offers the following distinctive traits:
| Distinctive Trait | Why It Matters | How It Works |
|---|---|---|
| 35B activated parameters | Efficient inference for massive 480B model | Uses 8 experts out of 160 total during inference, activating only 35B of 480B per forward pass |
| Native 256K token context | Repository-scale understanding in single pass | Processes 262,144 tokens natively, extended to 1M tokens with Yarn extrapolation for vast codebases |
| Specially designed function call format | Agentic coding requires precise tool invocation | Supports Qwen Code, CLINE, and other platforms with dedicated function call structure for tool use |
| Non-thinking mode only | Consistent behavior for coding tasks | Does not generate blocks, spec enable_thinking=False not required for predictable output |
Why It Exists
Qwen3 Coder 480B A35B was built to deliver autonomous coding agent capabilities that manage complex workflows beyond basic code generation, supporting interactive multi-turn reasoning and tool use across diverse programming ecosystems.
Benchmark Performance
Independent evaluations · Artificial Analysis
Accuracy & Capability Details
Compare Models Side-by-Side
Evaluate specifications, pricing, and independent benchmark indices
| Model Details | |||
|---|---|---|---|
| General Info | |||
| Provider | Alibaba | Anthropic | Anthropic |
| Release Date | July 22, 2025 | September 22, 2026 | September 28, 2026 |
| Knowledge Cutoff | Jun 2025 | - | Jun 2026 |
| Context & Limits | |||
| Context Window | 262K | 1M Best Context Window | 1M Best Context Window |
| Pricing (per 1M tokens) | |||
| Input Pricing | $0.30 Best Input Pricing | $4 | $2 |
| Output Pricing | $1.80 Best Output Pricing | $20 | $10 |
| Modalities | |||
| Inputs | text | textimagefile | textimagefile |
| Outputs | text | text | text |
| Benchmarks (0-100) | |||
| Intelligence Index | 11.9 | 57.6 Best Intelligence Index | 56.0 |
| Coding Index | - | - | - |
| Agentic Index | - | - | - |
Humanity's Last Exam
Extremely difficult logical reasoning and knowledge.
Long Context Reasoning
Logical reasoning over long context windows.
SciCode Benchmark
Scientific coding and mathematical modeling.
Independent evaluation data provided by Artificial Analysis. To view the latest benchmarks and full details, visit their official site.
Explore more from Alibaba
Other models by Alibaba
Qwen3.8 Max
Qwen3.8 2.4T A95B
Qwen3.8-Flash-Next
Qwen3.8 27B
Top AI Models
Leading alternatives by intelligence score
