Qwen3.5-122B-A10B

qwen/qwen3.5-122b-a10b
VisionToolsJSONReasoning
by Qwen · 2026-02-25

Qwen3.5 122B-A10B — open-weight MoE multimodal (text/image/video), 122B total / 10B active params, 32k context (vision mode).

ctx32.8K tokens
Max output65.5K
Inputtext + image + video
Outputtext
p50 TTFT4.00 s
INPUT$0.12/ 1M tokens
OUTPUT$0.92/ 1M tokens
p50 TTFT4.00 s7d
p95 TTFT4.46 s7d
TRAFFIC43.4Ktokens / 7d

Qwen3.5-122B-A10B is a large language model from the Qwen series by Alibaba Cloud. It uses a mixture-of-experts (MoE) architecture where only 10 billion parameters are activated per forward pass,…

What is Qwen3.5-122B-A10B?

Who is this model designed for?

What multimodal capabilities does it offer?

How does it compare to other Qwen models in the series?

Code samples

Call from any SDK

OpenAI-compatible — keep the SDK you already use

  • OpenAI SDKhttps://api.orcarouter.ai/v1
from openai import OpenAI

client = OpenAI(
    base_url="https://api.orcarouter.ai/v1",
    api_key="$ORCAROUTER_API_KEY",
)

response = client.chat.completions.create(
    model="qwen/qwen3.5-122b-a10b",
    messages=[{"role": "user", "content": "Hello"}],
)
print(response.choices[0].message.content)

Supported parameters

  • enable_search
  • enable_thinking
  • include_reasoning
  • logprobs
  • max_tokens
  • n
  • parallel_tool_calls
  • presence_penalty
  • reasoning
  • repetition_penalty
  • response_format
  • seed
  • stop
  • stream
  • stream_options
  • temperature
  • thinking_budget
  • tool_choice
  • tools
  • top_k
  • top_logprobs
  • top_p

Pricing

TierInput / 1M tokensOutput / 1M tokens
128K$0.115$0.917
256K$0.287$2.294
Tier selected by input token count of each request

Cost calculator

Tokens / month10MM
Input share70%%
Estimated / month $3.56

Estimate based on list price

Tiered pricing — this estimate uses base-tier rates.

Token & cost estimator

Input tokens: 20Cost per request: $0.000461

Estimate only — actual token counts depend on the provider's tokenizer.

Performance

p50 TTFT
4.00 s
Output speed
270 tok/s
p95 TTFT
4.46 s
Error rate
0%

Public benchmarks

45.7
AA Coding
Better than 52% of models compared
#59 of 123
32.3
AA Intelligence
Better than 40% of models compared
#75 of 125
GPQA Diamond
85.7
Humanity's Last Exam
23.4
IFBench
75.7
Long-Context Recall
66.7
SciCode
42.0
tau_banking
13.6
TerminalBench Hard
31.1
terminalbench_v2_1
47.6
τ²-Bench
93.6
Source: artificialanalysis.ai

How it compares

Qwen3.5-122B-A10Bqwen/qwen3-max-previewQwen3.5 397B A17Bqwen/qwen3.5-plus
Input $/M$0.12$0.86$0.17$0.12
Output $/M$0.92$3.44$1.03$0.69
Context33K262K33K1.0M
Quality7/108/108/108/10
Compare side-by-sideCompare side-by-sideCompare side-by-sideCompare side-by-side

FAQ

What is the cost of using Qwen3.5-122B-A10B on OrcaRouter?
Exact pricing is not publicly specified in the available facts. OrcaRouter typically charges per token for input and output, with multimodal tokens billed at a higher rate. Refer to OrcaRouter's pricing page or contact support for current rates.
What is the context window size of this model?
The context window supports up to 32,768 tokens (including messages, images, and video frames). The maximum output length is 65,536 tokens.
What are the main strengths of Qwen3.5-122B-A10B?
Its strengths include a high score on τ²-Bench (93.6) indicating strong tool-use and multi-step reasoning, multimodal input support (text, image, video), and a large output limit. The MoE architecture balances capacity and efficiency.
How does this model compare to Llama 3.1 or Claude?
Direct benchmarks are limited. This model has a strong tool-use score and multimodal input, while Llama 3.1 is text-only (except vision variants) and Claude has larger context windows. Performance will vary by task; testing on your specific data is recommended.
Does OrcaRouter handle user data and privacy when using this model?
Data handling policies are not detailed in the provided facts. Users should review OrcaRouter's terms of service and privacy policy. Typically, inputs and outputs may be processed on cloud servers; sensitive data should be treated accordingly.
How do I call this model using the OpenAI-compatible API?
Send a POST request to https://api.orcarouter.ai/v1/chat/completions with the model parameter set to 'qwen/qwen3.5-122b-a10b'. The API supports the same message format as OpenAI, including multimodal content via image_url and video blocks.
What is the maximum output token limit?
The model can generate up to 65,536 tokens in a single completion. This can be controlled with the max_tokens parameter in the API request.
Can this model process videos?
Yes, it accepts video input (along with text and images). The exact encoding method (e.g., frame sampling) is not specified, but the model can interpret video frames as part of the multimodal context.

Embed this badge

Qwen: Qwen3.5-122B-A10B$0.12/M in4000ms p50via OrcaRouter
HTML <a href="https://www.orcarouter.ai/models/qwen/qwen3.5-122b-a10b" target="_blank"> <img src="https://www.orcarouter.ai/embed/qwen/qwen3.5-122b-a10b.svg" alt="Qwen: Qwen3.5-122B-A10B on OrcaRouter" /> </a>
Markdown [![Qwen: Qwen3.5-122B-A10B](https://www.orcarouter.ai/embed/qwen/qwen3.5-122b-a10b.svg)](https://www.orcarouter.ai/models/qwen/qwen3.5-122b-a10b)

Model card as data

GET /api/public/models/qwen/qwen3.5-122b-a10bOpen
Machine-readable:/llms.txt/llms-full.txt