Qwen3.5-35B-A3B is a mid-sized Mixture-of-Experts (MoE) model with 35 billion total parameters, activating only 3 billion per token. The model comprises 40 layers with a hidden size of 2048 and utilizes tokenization with a notably large vocabulary size of 248,320. Its hybrid attention architecture combines Gated DeltaNet layers (linear attention) for fast processing of long sequences and Gated Attention layers (full attention) for precise contextual understanding. This enables the model to support a native context window of 262,144 tokens without quality degradation. Vision-language capabilities are integrated through early-fusion training, providing better image understanding compared to the Qwen3-VL series. The model supports two operational modes: Thinking for deep reasoning (mathematics, logic, code) and No-thinking for quick responses to simple queries. Inference is highly optimized; deploying the quantized format requires approximately 22–24 GB of GPU memory.
The model demonstrates impressive results on benchmarks, falling only slightly behind the flagship versions of the series. In language tests such as MMLU-Pro (85.3) and SuperGPQA (63.4), it outperforms larger models from the previous generation. Its agentic capabilities stand out in particular: the TAU2-Bench score (81.2) is the best in the family, indicating excellent proficiency in planning and executing multi-step tasks using tools. In multimodal analysis, it shows results close to top-tier models: MathVision (83.9), MMMU-Pro (75.1), OCRBench (91.0). It is important to note that this model forms the foundation of the Qwen3.5-Flash service.
The model's uniqueness lies in its versatility and efficiency, plus it distinguishes itself from previous versions with a significant leap in agent performance and multimodal understanding. This variant could be an excellent choice for companies developing sophisticated assistants, order processing systems, intelligent RAG systems with vast knowledge bases, and generally for any scenario requiring high-quality context understanding and generation while maintaining controlled, reasonable infrastructure costs.
| Model Name | Context | Type | GPU | TPS | Tooling | Status | Link |
|---|---|---|---|---|---|---|---|
| Qwen/Qwen3.5-35B-A3B-GPTQ-Int4 | 263K | Public | 2×4090 tensor |
227.0 | yes | AVAILABLE | chat |
curl https://chat.immers.cloud/v1/endpoints/generate/chat/completions \
-H "Content-Type: application/json" \
-H "Authorization: Bearer USER_API_KEY" \
--data-binary @- <<"EOF"
{"model": "qwen3.5-35b-a3b", "messages": [
{"role": "system", "content": "You are a helpful assistant."},
{"role": "user", "content": "Say this is a test"}
], "temperature": 0, "max_tokens": 150
}
EOF
$response = Invoke-WebRequest https://chat.immers.cloud/v1/endpoints/generate/chat/completions `
-Method POST `
-Headers @{
"Authorization" = "Bearer USER_API_KEY"
"Content-Type" = "application/json; charset=utf-8"
} `
-Body ([System.Text.Encoding]::UTF8.GetBytes((@{
model = "qwen3.5-35b-a3b"
messages = @(
@{ role = "system"; content = "You are a helpful assistant." },
@{ role = "user"; content = "Say this is a test" })
} | ConvertTo-Json -Depth 10)))
([System.Text.Encoding]::UTF8.GetString($response.RawContentStream.ToArray()) | ConvertFrom-Json).choices[0].message.content
#!pip install OpenAI --upgrade
from openai import OpenAI
client = OpenAI(
api_key="USER_API_KEY",
base_url="https://chat.immers.cloud/v1/endpoints/generate/",
)
chat_response = client.chat.completions.create(
model="qwen3.5-35b-a3b",
messages=[
{"role": "system", "content": "You are a helpful assistant."},
{"role": "user", "content": "Say this is a test"},
]
)
print(chat_response.choices[0].message.content)
# linux
export ANTHROPIC_BASE_URL=https://chat.immers.cloud/v1/endpoints/generate/
export ANTHROPIC_API_KEY=USER_API_KEY
export ANTHROPIC_MODEL=qwen3.5-35b-a3b
# windows (ps)
[Environment]::SetEnvironmentVariable("ANTHROPIC_BASE_URL", "https://chat.immers.cloud/v1/endpoints/generate/", "User")
[Environment]::SetEnvironmentVariable("ANTHROPIC_AUTH_TOKEN", "USER_API_KEY", "User")
[Environment]::SetEnvironmentVariable("ANTHROPIC_MODEL", "qwen3.5-35b-a3b", "User")
# linux
export OPENAI_BASE_URL=https://chat.immers.cloud/v1/endpoints/generate/
export OPENAI_API_KEY=USER_API_KEY
export OPENAI_MODEL=qwen3.5-35b-a3b
# windows (ps)
[Environment]::SetEnvironmentVariable("OPENAI_BASE_URL", "https://chat.immers.cloud/v1/endpoints/generate/", "User")
[Environment]::SetEnvironmentVariable("OPENAI_API_KEY", "USER_API_KEY", "User")
[Environment]::SetEnvironmentVariable("OPENAI_MODEL", "qwen3.5-35b-a3b", "User")
An endpoint is a server with pre-installed software that automatically downloads neural network weights and launches an OpenAI-compatible API to handle incoming requests.
A private endpoint is created within the user's project, belongs exclusively to the user, and is billed per server configuration in accordance with the infrastructure pricing.
We recommend creating private endpoints when you need to:
| Name | GPU | TPS | Max Concurrency | |||
|---|---|---|---|---|---|---|
262,144.0 tensor |
2 | $0.93 | 2.847 | Launch | ||
262,144.0 pipeline |
3 | $1.06 | 1.508 | Launch | ||
262,144.0 tensor |
4 | $1.26 | 2.060 | Launch | ||
262,144.0 tensor |
2 | $1.56 | 100.12 | 3.231 | Launch | |
262,144.0 tensor |
2 | $1.56 | 3.231 | Launch | ||
262,144.0 tensor |
2 | $1.92 | 227.00 | 3.213 | Launch | |
262,144.0 |
1 | $2.37 | 132.96 | 9.655 | Launch | |
262,144.0 tensor |
2 | $2.93 | 6.085 | Launch | ||
262,144.0 |
1 | $3.83 | 153.60 | 9.699 | Launch | |
262,144.0 |
1 | $4.11 | 189.85 | 12.227 | Launch | |
262,144.0 tensor |
2 | $4.61 | 23.610 | Launch | ||
262,144.0 |
1 | $4.74 | 20.667 | Launch | ||
262,144.0 tensor |
2 | $9.40 | 45.762 | Launch | ||
| Name | GPU | TPS | Max Concurrency | |||
|---|---|---|---|---|---|---|
262,144.0 pipeline |
3 | $1.34 | 2.446 | Launch | ||
262,144.0 tensor |
4 | $1.57 | 3.497 | Launch | ||
262,144.0 pipeline |
6 | $1.65 | 4.291 | Launch | ||
262,144.0 pipeline |
3 | $2.29 | 138.25 | 2.848 | Launch | |
262,144.0 |
1 | $2.37 | 154.74 | 6.909 | Launch | |
262,144.0 pipeline |
3 | $2.83 | 130.33 | 2.918 | Launch | |
262,144.0 tensor |
4 | $2.89 | 3.886 | Launch | ||
262,144.0 tensor |
2 | $2.93 | 2.936 | Launch | ||
262,144.0 tensor |
4 | $3.01 | 3.886 | Launch | ||
262,144.0 tensor |
4 | $3.60 | 3.872 | Launch | ||
262,144.0 |
1 | $3.83 | 165.39 | 7.058 | Launch | |
262,144.0 |
1 | $4.11 | 8.926 | Launch | ||
262,144.0 tensor |
2 | $4.61 | 20.460 | Launch | ||
262,144.0 |
1 | $4.74 | 17.421 | Launch | ||
262,144.0 tensor |
2 | $9.40 | 42.613 | Launch | ||
| Name | GPU | TPS | Max Concurrency | |||
|---|---|---|---|---|---|---|
262,144.0 |
1 | $4.12 | 180.34 | 3.456 | Launch | |
262,144.0 pipeline |
3 | $4.35 | 1.456 | Launch | ||
262,144.0 |
1 | $4.74 | 11.886 | Launch | ||
262,144.0 tensor |
2 | $4.93 | 114.38 | 14.787 | Launch | |
262,144.0 tensor |
2 | $4.94 | 14.787 | Launch | ||
262,144.0 tensor |
4 | $5.76 | 3.794 | Launch | ||
262,144.0 pipeline |
6 | $5.84 | 7.127 | Launch | ||
262,144.0 tensor |
8 | $6.04 | 4.208 | Launch | ||
262,144.0 tensor |
8 | $7.52 | 4.193 | Launch | ||
262,144.0 tensor |
2 | $7.85 | 14.762 | Launch | ||
262,144.0 tensor |
2 | $9.41 | 36.940 | Launch | ||
Contact our dedicated neural networks support team at support@immers.cloud or send your request to the sales department at sale@immers.cloud.