Qwen 3 Coder 480B A35B
Language modelActive
Qwen 3 Coder 480B A35B
Alibaba
- Released
- July 23, 2025
- Data date
- September 8, 2026
- Speed
- 58.6 tokens/sOutput throughput
- Price
- Input · Output / 1M tokens
- Input
- Text
- Output
- Text
Qwen 3 Coder 480B A35B is Alibaba's large open model for agentic coding. It activates 35 billion of 480 billion parameters and runs without a Thinking mode, while still supporting tools and structured output. This Coder configuration does not accept multimodal input.
The model card lists 262,144 native context tokens and extension to 1 million with YaRN. Code made up 70% of its 7.5 trillion training tokens. For fewer total parameters, see Qwen 3 Coder 30B A3B.
Specifications and access
| Specification | Value and source |
|---|---|
| Weights | Open weightsSource |
| Access | API, LocalSource |
| Parameters | 480BSource |
| Active parameters | 35BSource |
| Context window | 262,144 tokensSource |
| Weights repository | Qwen/Qwen3-Coder-480B-A35B-InstructSource |
| License | Apache-2.0Source |
| Architecture | Qwen3 sparse MoE causal language model optimized for agentic codingSource |
| Input | TextSource |
| Output | TextSource |
| Reasoning mode | non-thinking onlySource |
| Tool calling | YesSource |
| Structured output | YesSource |
| Native context window | 262,144 tokensSource |
| Maximum context window | 1,000,000 tokensSource |
| Parameters | 480 BSource |
| Active parameters | 35 BSource |
| Layers | 62Source |
| Total experts | 160Source |
| Active experts per token | |
| Training tokens | 7.5 trillionSource |
| Code share of training data | 70%Source |
| Repository created | 2025-07-22 (Repository creation date is a precise publication proxy, but may differ from the public announcement date or an individual checkpoint's first availability.)Source |
Page 1 of 2
Measurements without matching peer values
These measurements have no matching peer values under the same test conditions. Their original values and sources remain available here.
Terminal-Bench (Show measurement, test conditions, and source)
- Source value
- 37.5
- Score
- 37.5
- Metric
- percent
- Unit
- %
- Category
- coding
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- Qwen3-Coder official release benchmark table
- Status
- active
- Retrieved at
- 2026-09-08
- Methodology
- Vendor-reported result. Qwen3-Coder-480B-A35B-Instruct column; official Agentic Coding table
SWE-bench Verified (OpenHands, 500 turns) (Show measurement, test conditions, and source)
- Source value
- 69.6
- Score
- 69.6
- Metric
- percent
- Unit
- %
- Category
- coding
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- Qwen3-Coder official release benchmark table
- Status
- active
- Retrieved at
- 2026-09-08
- Methodology
- Vendor-reported result. Qwen3-Coder-480B-A35B-Instruct column; SWE-bench Verified with OpenHands, 500 turns
SWE-bench Verified (OpenHands, 100 turns) (Show measurement, test conditions, and source)
- Source value
- 67
- Score
- 67
- Metric
- percent
- Unit
- %
- Category
- coding
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- Qwen3-Coder official release benchmark table
- Status
- active
- Retrieved at
- 2026-09-08
- Methodology
- Vendor-reported result. Qwen3-Coder-480B-A35B-Instruct column; SWE-bench Verified with OpenHands, 100 turns
SWE-bench Live (Show measurement, test conditions, and source)
- Source value
- 26.3
- Score
- 26.3
- Metric
- percent
- Unit
- %
- Category
- coding
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- Qwen3-Coder official release benchmark table
- Status
- active
- Retrieved at
- 2026-09-08
- Methodology
- Vendor-reported result. Qwen3-Coder-480B-A35B-Instruct column; official Agentic Coding table
SWE-bench Multilingual (Show measurement, test conditions, and source)
- Source value
- 54.7
- Score
- 54.7
- Metric
- percent
- Unit
- %
- Category
- coding
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- Qwen3-Coder official release benchmark table
- Status
- active
- Retrieved at
- 2026-09-08
- Methodology
- Vendor-reported result. Qwen3-Coder-480B-A35B-Instruct column; official Agentic Coding table
Multi-SWE-bench mini (Show measurement, test conditions, and source)
- Source value
- 25.8
- Score
- 25.8
- Metric
- percent
- Unit
- %
- Category
- coding
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- Qwen3-Coder official release benchmark table
- Status
- active
- Retrieved at
- 2026-09-08
- Methodology
- Vendor-reported result. Qwen3-Coder-480B-A35B-Instruct column; official Agentic Coding table
Multi-SWE-bench flash (Show measurement, test conditions, and source)
- Source value
- 27
- Score
- 27
- Metric
- percent
- Unit
- %
- Category
- coding
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- Qwen3-Coder official release benchmark table
- Status
- active
- Retrieved at
- 2026-09-08
- Methodology
- Vendor-reported result. Qwen3-Coder-480B-A35B-Instruct column; official Agentic Coding table
Aider-Polyglot (Show measurement, test conditions, and source)
- Source value
- 61.8
- Score
- 61.8
- Metric
- percent
- Unit
- %
- Category
- coding
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- Qwen3-Coder official release benchmark table
- Status
- active
- Retrieved at
- 2026-09-08
- Methodology
- Vendor-reported result. Qwen3-Coder-480B-A35B-Instruct column; official Agentic Coding table
Spider2 (Show measurement, test conditions, and source)
- Source value
- 31.1
- Score
- 31.1
- Metric
- percent
- Unit
- %
- Category
- coding
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- Qwen3-Coder official release benchmark table
- Status
- active
- Retrieved at
- 2026-09-08
- Methodology
- Vendor-reported result. Qwen3-Coder-480B-A35B-Instruct column; official Agentic Coding table
WebArena (Show measurement, test conditions, and source)
- Source value
- 49.9
- Score
- 49.9
- Metric
- percent
- Unit
- %
- Category
- browser-use
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- Qwen3-Coder official release benchmark table
- Status
- active
- Retrieved at
- 2026-09-08
- Methodology
- Vendor-reported result. Qwen3-Coder-480B-A35B-Instruct column; official Agentic Browser Use table
Mind2Web (Show measurement, test conditions, and source)
- Source value
- 55.8
- Score
- 55.8
- Metric
- percent
- Unit
- %
- Category
- browser-use
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- Qwen3-Coder official release benchmark table
- Status
- active
- Retrieved at
- 2026-09-08
- Methodology
- Vendor-reported result. Qwen3-Coder-480B-A35B-Instruct column; official Agentic Browser Use table
BFCL-v3 (Show measurement, test conditions, and source)
- Source value
- 68.7
- Score
- 68.7
- Metric
- percent
- Unit
- %
- Category
- tool-use
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- Qwen3-Coder official release benchmark table
- Status
- active
- Retrieved at
- 2026-09-08
- Methodology
- Vendor-reported result. Qwen3-Coder-480B-A35B-Instruct column; official Agentic Tool Use table
Page 1 of 2
Our tests
SVG short test
One SVG prompt and fixed output settings. Captures are shown only with 0 reported reasoning tokens or documented lack of reasoning controls. Identical test conditions do not make the models' internal compute identical.
Earlier tests
Pelican on a bicycle
A community classic for free-form SVG drawing.
Qwen 3 Coder 480B A35B
Reasoning (requested): not specified
Current attempt
- Run date
- Oct 1, 2026
- API provider
- openrouter
- Requested API model
- qwen/qwen3-coder
- API model reported in response
- qwen/qwen3-coder
- Reasoning setting
- not specified
- Pinned endpoint
- Alibaba | qwen/qwen3-coder-480b-a35b-07-25
- Temperature
- Provider default
- Token limit including reasoning
- 8,192 tokens
- Test protocol
- community-visual-tests-v1
- Technical checks
- Passed, not a quality rating
- Response time
- 10.2 s
Visually reviewed. This individual test does not establish overall model quality.
Prompt and test conditions
Generate an SVG of a pelican riding a bicycle
Token limit including reasoning: 8,192 tokens
Task origin (Simon Willison)Popularity
42.6 / 100
Sources and data date
| Type | Evidence and data date |
|---|---|
| Speed | Alibaba. Output speed after the first streaming chunk, not total response time. Classified as deprecated by Artificial Analysis. Artificial Analysis maintains only the default 10K-input workload there. Retrieved September 8, 2026. |
| Price | Base API prices without caching or batch discounts. Higher context tiers and other rates are listed under Costs. |
| Input | Documented modalities. Retrieved September 8, 2026. |
| Output | Documented modalities. Retrieved September 8, 2026. |
| Research status | Research date September 8, 2026. 4 source URLs checked. This documents the inspected sources, not an exhaustive inventory of every publication. |
| Not found in the inspected sources | knowledge-cutoff Original wording No exact training-data cutoff was stated in the checked official card/report. |
| Additional source | Qwen3-Coder official release benchmark table (retrieved September 8, 2026) · Qwen3-Coder official release · Editorial description reviewed October 2, 2026 |
| Additional source | Model metadata (retrieved September 8, 2026) |
| Additional source | OpenRouter: Popularity index (derived), CC BY 4.0 |
| Additional source | Vercel AI Gateway: Popularity index (derived), CC BY 4.0 |
| Additional source | Creative Commons Attribution 4.0 |