impulse

Qwen3.6 35B A3B

Open weights

Qwen3.6-35B-A3B is an open-weight multimodal model from Alibaba Cloud with 35 billion total parameters and 3 billion active parameters per token.

API id Qwen3.6-35B-A3BReleased 27 Apr 2026 (Apr 2026)
Input / 1M
$0.10
at W&B Inference
Output / 1M
$0.89
at W&B Inference
Context window
262K
tokens
Max output
262K
tokens per reply
Hosts
12
12 listings
Released
Apr 2026
27 Apr 2026

Pricing

List prices at W&B Inference, in US dollars per million tokens.

TierInput / 1MOutput / 1M
Standard$0.10$0.89
Cached input
65% off input
$0.035

What it costs a month

At W&B Inference list prices, 30 days.

  • Chat assistant5,000 requests a day, 2K in / 0.5K out, cached$90.90
  • RAG over documents2,000 requests a day, 8K in / 0.6K out, cached$64.44
  • Coding agent1,000 requests a day, 40K in / 2K out, cached$119
  • Overnight batch20,000 requests a day, 4K in / 0.4K out$454
Use your own numbers

Where to run it

12 hosts serve Qwen3.6 35B A3B. The comparison is on blended price, 3 parts input to 1 part output.

HostIn / out per 1M
SailCloud
Qwen3.6-35B-A3B
$0.05 / $0.40
−54% vs headline
qwen3-6-35b-a3b
$0.07 / $0.42
−47% vs headline
qwen3.6-35b-a3b
$0.14 / $0.45
−27% vs headline
qwen3.6-35b-a3b
$0.15 / $0.50
−20% vs headline
W&B InferenceInference
Qwen3.6-35B-A3B
$0.10 / $0.89
DeepInfraInference
Qwen3.6-35B-A3B
$0.10 / $0.95
+5% vs headline
ZeroGPUCloud
qwen3.6-35b-a3b
$0.20 / $1.50
+76% vs headline
GMI CloudInference
Qwen3.6-35B-A3B
$0.248 / $1.49
+87% vs headline
Novita AIInference
qwen3.6-35b-a3b
$0.248 / $1.49
+87% vs headline
qwen3.6-35b-a3b
$0.25 / $1.50
+89% vs headline
AIHubMixGateway
qwen3.6-35b-a3b
$0.254 / $1.52
+92% vs headline
qwen3.6-35b-a3b
$0.375 / $2.25
+184% vs headline

Capabilities

Core

  • Streaming
  • Function calling
  • JSON mode
  • Structured outputs

Multimodal

  • Vision
  • Image input
  • Audio
  • Image generation
  • Video

Advanced

  • Reasoning
  • Tool calling
  • MCP
  • Prompt caching
  • Batch API
  • Embeddings
Takes in
TextImages
Produces
Text
Features
Prompt caching
Open weights
License: Apache-2.0
Runs in about 86.3 GB of GPU memory
Runtimes: Ollama, SGLang, llama.cpp, vLLM

Cheaper alternatives

Current models that cost less and keep most of what this one can do: tools, vision, reasoning, structured output, caching and enough context.

Same model, cheaper host

  • Sail
    Qwen3.6-35B-A3B
    $0.05 / $0.40
    54% less
  • EmpirioLabs AI
    qwen3-6-35b-a3b
    $0.07 / $0.42
    47% less
  • Pinstripes
    qwen3.6-35b-a3b
    $0.14 / $0.45
    27% less
  • LibertAI
    qwen3.6-35b-a3b
    $0.15 / $0.50
    20% less

Smaller Alibaba Qwen models

No cheaper Alibaba Qwen model keeps its key capabilities.

Other frontier labs

  • Claude Haiku 5.5
    Anthropic · on Anthropic
    $0.10 / $0.50
    Has every key capability
    33% less
  • GPT-6 Luna
    OpenAI · on OpenAI
    $0.10 / $0.50
    Has every key capability
    33% less
  • GPT-4o Mini
    OpenAI · on OpenAI
    $0.15 / $0.60
    Lacks reasoning
    12% less

Open-weight picks

  • Deepseekv4.1 Flash
    DeepSeek · on Sarvam
    $0.149 / $0.585
    Has every key capability
    13% less
  • Qwen3.8 27B
    Alibaba Qwen · on EmpirioLabs AI
    $0.17 / $0.50
    Has every key capability
    15% less
  • Nemotron 3.5 Super VL
    NVIDIA · on Kyma API
    $0.081 / $0.472
    Has every key capability
    40% less
  • MiMo V2.6 Flash
    Xiaomi · on GMI Cloud
    $0.112 / $0.224
    Has every key capability
    53% less

Model line

Newer and older versions, at each lab's own price.

  1. Apr 2026Qwen3.6 35B A3BLatest$0.10 / $0.89
  2. Feb 2026Qwen3.5 35B A3B$0.14 / $1.00

Call it

Using W&B Inference's API.

Python
from litellm import completion

response = completion(
    model="wandb/Qwen3.6-35B-A3B",
    messages=[{"role": "user", "content": "Hello"}],
)
print(response.choices[0].message.content)

Quick answers

How much does Qwen3.6 35B A3B cost?

On W&B Inference, Qwen3.6 35B A3B costs $0.10 per million input tokens and $0.89 per million output tokens. Cached input costs $0.035 per million tokens. The cheapest host is Sail, at $0.05 per million input tokens and $0.40 per million output tokens.

Which APIs offer Qwen3.6 35B A3B?

Qwen3.6 35B A3B runs on 12 APIs: Sail, EmpirioLabs AI, Pinstripes, LibertAI, W&B Inference, DeepInfra, ZeroGPU, GMI Cloud, Novita AI and 3 more.

What is Qwen3.6 35B A3B's context window?

262K tokens, with up to 262K tokens in one reply.

Is Qwen3.6 35B A3B being retired?

No retirement date is scheduled for it on any API we track.

Is there a newer version of Qwen3.6 35B A3B?

No. Qwen3.6 35B A3B is the newest model in its line.

What are cheaper alternatives to Qwen3.6 35B A3B?

Models that cost less and keep most of what it can do: Claude Haiku 5.5 on Anthropic (33% less), GPT-6 Luna on OpenAI (33% less) and Deepseekv4.1 Flash on Sarvam (13% less). Qwen3.6 35B A3B itself costs 54% less on Sail. Savings are on a mix of 3 input tokens to 1 output token.

When was Qwen3.6 35B A3B released?

Alibaba Qwen released it on 27 Apr 2026.

What is Qwen3.6 35B A3B's API model id?

Use Qwen3.6-35B-A3B with W&B Inference. Other hosts use their own ids: qwen3-6-35b-a3b on EmpirioLabs AI, qwen3.6-35b-a3b on Pinstripes, qwen3.6-35b-a3b on LibertAI, qwen3.6-35b-a3b on ZeroGPU.

Sources

Raw catalog entry for Qwen3.6-35B-A3B on W&B Inference
{
  "cache_read_input_token_cost": 3.5e-8,
  "supports_reasoning": true,
  "max_output_tokens": 262144,
  "max_input_tokens": 262000,
  "input_cost_per_token": 1e-7,
  "output_cost_per_token": 8.9e-7,
  "litellm_provider": "wandb",
  "mode": "chat",
  "supports_vision": true,
  "source": "https://www.coreweave.com/products/coreweave-forge/serverless-inference",
  "supports_prompt_caching": true
}