Qwen3.6 35B-A3B: $0.8438 blended vs. $0.85 for Gemini 3.5 Flash Lite

2026-10-21 · news · generated from AI Signals data, numbers verified automatically

Alibaba (Qwen) raised pricing for Qwen3.6 35B-A3B on October 20, 2026, bringing its listed blended rate to $0.8438 from $0.5573, a 51.4% increase. That leaves it just below Gemini 3.5 Flash Lite’s $0.85 blended rate. For developers choosing an API, however, the input and output rates matter more than that narrow blended-price gap: the supplied monthly scenarios put Qwen below Gemini for the coding-agent workload, but above it for chatbot and RAG workloads.

What changed

Qwen3.6 35B-A3B’s input price increased to $0.375 per million tokens from $0.248. Its output price rose to $2.25 per million tokens from $1.485.

The context window remains unchanged at 262,144 tokens. The model lists a maximum output of 65,536 tokens and supports reasoning, tool calling, and vision. It has open weights and a release date of April 17, 2026.

This update is a pricing change, not a context-window expansion. Existing users assessing the impact should distinguish the higher token rates from the unchanged context allowance. The listed blended rate provides a comparison point, but the workload examples show why it is not a substitute for pricing an application’s actual input and output volume.

What it costs in practice

The following are monthly costs at base-tier list prices. Input and output volumes are in millions of tokens per month.

Scenario Input tokens (M) Output tokens (M) Qwen3.6 35B-A3B Gemini 2.5 Flash Gemini 3.5 Flash Lite GLM-4.5V
Chatbot 10 2 $8.25 $8 $8 $9.60
RAG app 50 5 $30 $27.50 $27.50 $39
Coding agent 20 10 $30 $31 $31 $30

The chatbot and RAG examples favor the Gemini models on cost, despite Qwen’s lower listed blended rate. The coding-agent example reverses that ordering: Qwen costs $30, compared with $31 for either Gemini model, and matches GLM-4.5V.

These scenarios make the input/output split important. Qwen’s input rate is higher than the Gemini rates, while its output rate is lower.

How it compares

Prices below are per million tokens. Context and maximum output are measured in tokens.

Competitor Input Output Blended Context Maximum output Open weights
Gemini 2.5 Flash $0.30 $2.50 $0.85 1,048,576 65,536 No
Gemini 3.5 Flash Lite $0.30 $2.50 $0.85 1,048,576 65,536 No
GLM-4.5V $0.60 $1.80 $0.90 64,000 16,384 Yes

All listed competitors support reasoning, tool calling, and vision. The Gemini models offer larger context windows than Qwen and the same maximum output. GLM-4.5V has a smaller context window and maximum output, with a higher input rate but a lower output rate than Qwen.

Who should care

Existing Qwen API users should revisit budgets following the 51.4% blended-price increase. Developers selecting a model should compare token mix alongside context limits and open-weight availability.

Qwen is not the cheapest model in its class. Its listed blended rate remains below those of the competitors shown, but the monthly examples demonstrate that a lower blended rate does not guarantee a lower application bill. These comparisons establish pricing and listed capabilities, not relative model quality.

Qwen3.6 35B-A3B: live pricing →