VENICE ATLAS · MODEL DOSSIER · QWEN 3.6 35B A3B

Model overview

Qwen 3.6 35B A3B

Name: Qwen 3.6 35B A3B
Brand: venice.ai
Price: 0.1000 USD
Availability: InStock

Qwen 3.6 35B A3B is a fast mixture-of-experts model with 35B total parameters and ~3B active per token. Strong at agentic coding, STEM reasoning, and tool use, with a native 256K context window.

qwen3-6-35b-a3b · venice.ai

Text · Venice catalog$0.10 per 1M tokensPrice on Venice AI

ModalityTextOn Venice AI

Primary price$0.10 per 1M tokensper 1M tokens

Context window256KMax tokens

Benchmarks49Benchmark results

Benchmark results

See the tests behind each score. They use the benchmark’s original scoring scale.

Economics#816.28 rating Finance#1429.31 rating Frontend Development#2417.66 rating Grounding#815.06 rating Healthcare#1033.37 rating Image To Text#1114.33 rating Language#1729.17 rating Legal#1129.31 rating Long Context#2118.24 rating Spatial Reasoning#2023.39 rating Structured Output#726.55 rating Text To Image#219.80 rating Video understanding#622.33 rating

Agents5 results

DeepPlanningdeep-planning · agents

25.9%

ReportedOct 2026

MCP-Markmcp-mark · agents

37.0%

ReportedOct 2026

TAU3-Benchtau3-bench · agents

67.2%

ReportedOct 2026

Toolathlontoolathlon · agents

26.9%

ReportedOct 2026

VITA-Benchvita-bench · agents

35.6%

ReportedOct 2026

Coding4 results

Claw-Evalclaw-eval · code

50.0%

ReportedOct 2026

NL2Reponl2repo · code

29.4%

ReportedOct 2026

SkillsBenchskillsbench · code

28.7%

ReportedOct 2026

ZClawBenchzclawbench · code

52.6%

ReportedOct 2026

General intelligence1 result

C-Evalc-eval · general

90.0%

ReportedOct 2026

Language2 results

MMLU-Prommlu-pro · language

85.2%

ReportedOct 2026

MMLU-Reduxmmlu-redux · language

93.3%

ReportedOct 2026

Math6 results

AIME 2026aime-2026 · math

92.7%

ReportedOct 2026

HMMT 2025hmmt-2025 · math

90.7%

ReportedOct 2026

HMMT Feb 26hmmt-feb-26 · math

83.6%

ReportedOct 2026

HMMT25hmmt25 · math

89.1%

ReportedOct 2026

IMO-AnswerBenchimo-answerbench · math

78.9%

ReportedOct 2026

MathVista-Minimathvista-mini · math

86.4%

ReportedOct 2026

Multimodal3 results

MMMUmmmu · multimodal

81.7%

ReportedOct 2026

VideoMME w/o sub.videomme-w-o-sub. · multimodal

82.5%

ReportedOct 2026

VideoMMMUvideommmu · multimodal

83.7%

ReportedOct 2026

Reasoning9 results

CharXiv-Rcharxiv-r · reasoning

78.0%

ReportedOct 2026

GPQAgpqa · reasoning

86.0%

ReportedOct 2026

Humanity's Last Examhumanity's-last-exam · reasoning

21.4%

ReportedOct 2026

LiveCodeBench v6livecodebench-v6 · reasoning

80.4%

ReportedOct 2026

MCP Atlasmcp-atlas · reasoning

62.8%

ReportedOct 2026

SuperGPQAsupergpqa · reasoning

64.7%

ReportedOct 2026

SWE-bench Multilingualswe-bench-multilingual · reasoning

67.2%

ReportedOct 2026

SWE-Bench Proswe-bench-pro · reasoning

49.5%

ReportedOct 2026

SWE-Bench Verifiedswe-bench-verified · reasoning

73.4%

ReportedOct 2026

Learn what each benchmark measures ↗

Related models

Qwen 3.6 35B A3B

Benchmark results

More from qwen