Qwen3.6-35B-A3B

Model IntroductionText

Qwen3.6-35B-A3B is a 35B-parameter (3B activated) sparse MoE release of the Qwen3.6 series with a vision encoder, tuned for stability and real-world coding utility.

Introduction

Following the February release of the Qwen3.5 series, we're pleased to share the first open-weight variant of Qwen3.6. Built on direct feedback from the community, Qwen3.6 prioritizes stability and real-world utility, offering developers a more intuitive, responsive, and genuinely productive coding experience.

Key Features

  • Agentic Coding: the model now handles frontend workflows and repository-level reasoning with greater fluency and precision.
  • Thinking Preservation: we've introduced a new option to retain reasoning context from historical messages, streamlining iterative development and reducing overhead.

Benchmarks

Benchmark (Area)Score*
SWE-bench Verified (Coding Agent)73.4 (Eval)
SWE-bench Multilingual (Coding Agent)67.2 (Eval)
SWE-bench Pro (Coding Agent)49.5 (Eval)
Terminal-Bench 2.0 (Coding Agent)51.5 (Eval)
Claw-Eval Avg (Coding Agent)68.7 (Eval)
Claw-Eval Pass^3 (Coding Agent)50.0 (Eval)
SkillsBench Avg5 (Coding Agent)28.7 (Eval)
QwenClawBench (Coding Agent)52.6 (Eval)
NL2Repo (Coding Agent)29.4 (Eval)
QwenWebBench (Coding Agent)1397 (Eval)
TAU3-Bench (General Agent)67.2 (Eval)
VITA-Bench (General Agent)35.6 (Eval)
DeepPlanning (General Agent)25.9 (Eval)
Tool Decathlon (General Agent)26.9 (Eval)
MCPMark (General Agent)37.0 (Eval)
MCP-Atlas (General Agent)62.8 (Eval)
WideSearch (General Agent)60.1 (Eval)
MMLU-Pro (Knowledge)85.2 (Eval)
MMLU-Redux (Knowledge)93.3 (Eval)
SuperGPQA (Knowledge)64.7 (Eval)
C-Eval (Knowledge)90.0 (Eval)
GPQA (STEM & Reasoning)86.0 (Eval)
HLE (STEM & Reasoning)21.4 (Eval)
LiveCodeBench v6 (STEM & Reasoning)80.4 (Eval)
HMMT Feb 25 (STEM & Reasoning)90.7 (Eval)
HMMT Nov 25 (STEM & Reasoning)89.1 (Eval)
HMMT Feb 26 (STEM & Reasoning)83.6 (Eval)
IMOAnswerBench (STEM & Reasoning)78.9 (Eval)
AIME26 (STEM & Reasoning)92.7 (Eval)
MMMU (STEM and Puzzle)81.7 (Eval)
MMMU-Pro (STEM and Puzzle)75.3 (Eval)
Mathvista(mini) (STEM and Puzzle)86.4 (Eval)
ZEROBench_sub (STEM and Puzzle)34.4 (Eval)
RealWorldQA (General VQA)85.3 (Eval)
MMBench EN-DEV-v1.1 (General VQA)92.8 (Eval)
SimpleVQA (General VQA)58.9 (Eval)
HallusionBench (General VQA)69.8 (Eval)
OmniDocBench1.5 (Text Recognition and Document Understanding)89.9 (Eval)
CharXiv(RQ) (Text Recognition and Document Understanding)78.0 (Eval)
CC-OCR (Text Recognition and Document Understanding)81.9 (Eval)
AI2D_TEST (Text Recognition and Document Understanding)92.7 (Eval)
RefCOCO(avg) (Spatial Intelligence)92.0 (Eval)
ODInW13 (Spatial Intelligence)50.8 (Eval)
EmbSpatialBench (Spatial Intelligence)84.3 (Eval)
RefSpatialBench (Spatial Intelligence)64.3 (Eval)
VideoMME (w sub.) (Video Understanding)86.6 (Eval)
VideoMME (w/o sub.) (Video Understanding)82.5 (Eval)
VideoMMMU (Video Understanding)83.7 (Eval)
MLVU (Video Understanding)86.2 (Eval)
MVBench (Video Understanding)74.6 (Eval)
LVBench (Video Understanding)71.4 (Eval)

API request example

Endpoint /v1/chat/completions. Replace <API_KEY> with a key from the console.

Standard OpenAI Chat Completions. Swap `model` for the target slug.

curl https://api.ravenrouter.ai/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer <API_KEY>" \
  -d '{
  "model": "Qwen3.6-35B-A3B",
  "messages": [
    {
      "role": "user",
      "content": "Hello! Introduce yourself in one sentence."
    }
  ],
  "max_tokens": 64
}'

More models: API Reference → Examples by model

Qwen3.6-35B-A3B API — $0.29/1M input, $1.74/1M output | RavenRouter AI