Qwen3 Coder 30B A3B Instruct
by AlibabaRank #370Score 40.0
Qwen3-Coder-30B-A3B-Instruct is a 30.5B parameter Mixture-of-Experts (MoE) model with 128 experts (8 active per forward pass), designed for advanced code generation, repository-scale understanding, and agentic tool use. Built on the...
Performance Overview
Score
40.0
Rank
#370
24h Change
0
7d Change
0
State
stableConfidence
highSignal Scores
| Signal | Normalized | Weight | Contribution | Freshness |
|---|---|---|---|---|
Capabilities capability | 50.0 | 30% | 15.0 | 2026-09-22T14:24:10.670Z |
Pricing pricing_tier | 99.7 | 25% | 24.9 | 2026-09-22T14:24:10.670Z |
Context Window context_window | 86.0 | 15% | 12.9 | 2026-09-22T14:24:10.670Z |
Recency recency | 56.7 | 15% | 8.5 | 2026-09-22T14:24:10.670Z |
Output Capacity output_capacity | 85.9 | 15% | 12.9 | 2026-09-22T14:24:10.670Z |
Top Drivers
positive
Pricing
$0.28/M output tokens
$0.28
neutral
Capabilities
Supports tools, JSON mode, streaming
3/7
positive
Context Window
262K token context window
262K
positive
Output Capacity
Up to 236K output tokens per request
236K
Capabilities
| Capability | Supported |
|---|---|
| Vision | No |
| Reasoning | No |
| JSON Mode | Yes |
| Streaming | Yes |
| Function Calling | Yes |
| Web Search | No |
Pricing
Input / 1M tokens
$0.07
Output / 1M tokens
$0.28
Context Window
262K
Max Output
236K