Qwen3.5-397B-A17B vs Qwen3.5-35B-A3B Comparison

Comparing Qwen3.5-397B-A17B and Qwen3.5-35B-A3B across benchmarks, pricing, and capabilities.

Want to compare interactively?Try the playground

Performance Benchmarks

Comparative analysis across standard metrics

32 benchmarks

Qwen3.5-397B-A17B outperforms in 31 benchmarks (AA-LCR, BFCL-V4, BrowseComp, BrowseComp-zh, C-Eval, DeepPlanning, Global PIQA, GPQA, HMMT 2025, HMMT25, IFBench, IFEval, Include, LiveCodeBench v6, LongBench v2, MAXIFE, MMLU-Pro, MMLU-ProX, MMLU-Redux, MMMLU, Multi-Challenge, NOVA-63, PolyMATH, Seal-0, SuperGPQA, SWE-Bench Verified, t2-bench, Terminal-Bench 2.0, VITA-Bench, WideSearch, WMT24++), while Qwen3.5-35B-A3B is better at 1 benchmark (Humanity's Last Exam).

Qwen3.5-397B-A17B significantly outperforms across most benchmarks.

Tue Mar 17 2026 • llm-stats.com

Arena Performance

Human preference votes

Pricing Analysis

Price comparison per million tokens

Qwen3.5-35B-A3B costs less

For input processing, Qwen3.5-397B-A17B ($0.60/1M tokens) is 2.4x more expensive than Qwen3.5-35B-A3B ($0.25/1M tokens).

For output processing, Qwen3.5-397B-A17B ($3.60/1M tokens) is 1.8x more expensive than Qwen3.5-35B-A3B ($2.00/1M tokens).

In conclusion, Qwen3.5-397B-A17B is more expensive than Qwen3.5-35B-A3B.*

* Using a 3:1 ratio of input to output tokens

Lowest available price from all providers

Tue Mar 17 2026 • llm-stats.com

Qwen3.5-397B-A17B

Input tokens$0.60

Output tokens$3.60

Best providerNovita

Qwen3.5-35B-A3B

Input tokens$0.25

Output tokens$2.00

Best providerNovita

Notice missing or incorrect data?Start an Issue→

Model Size

Parameter count comparison

362.0B diff

Qwen3.5-397B-A17B has 362.0B more parameters than Qwen3.5-35B-A3B, making it 1034.3% larger.

Qwen3.5-397B-A17B

397.0Bparameters

Qwen3.5-35B-A3B

35.0Bparameters

397.0B

Qwen3.5-397B-A17B

35.0B

Qwen3.5-35B-A3B

Context Window

Maximum input and output token capacity

Both models have the same input context window of 262,144 tokens. Qwen3.5-35B-A3B can generate longer responses up to 65,000 tokens, while Qwen3.5-397B-A17B is limited to 64,000 tokens.

Qwen3.5-397B-A17B

Input262,144 tokens

Output64,000 tokens

Qwen3.5-35B-A3B

Input262,144 tokens

Output65,000 tokens

Tue Mar 17 2026 • llm-stats.com

Input Capabilities

Supported data types and modalities

Both Qwen3.5-397B-A17B and Qwen3.5-35B-A3B support multimodal inputs.

They are both capable of processing various types of data, offering versatility in application.

Qwen3.5-397B-A17B

Text

Images

Audio

Video

Qwen3.5-35B-A3B

Text

Images

Audio

Video

License

Usage and distribution terms

Both models are licensed under Apache 2.0.

Both models share the same licensing terms, providing consistent usage rights.

Qwen3.5-397B-A17B

Apache 2.0

Open weights

Qwen3.5-35B-A3B

Apache 2.0

Open weights

Release Timeline

When each model was launched

Qwen3.5-397B-A17B was released on 2026-02-16, while Qwen3.5-35B-A3B was released on 2026-02-24.

Qwen3.5-35B-A3B is 0 month newer than Qwen3.5-397B-A17B.

Qwen3.5-397B-A17B

Feb 16, 2026

4 weeks ago

Qwen3.5-35B-A3B

Feb 24, 2026

3 weeks ago

1w newer

Knowledge Cutoff

When training data ends

Neither model specifies a knowledge cutoff date.

Unable to compare the recency of their training data.

No cutoff dates available

Provider Availability

Qwen3.5-397B-A17B is available from Novita. Qwen3.5-35B-A3B is available from Novita. The availability of providers can affect quality of the model and reliability.

Qwen3.5-397B-A17B

Novita

Input Price:Input: $0.60/1MOutput Price:Output: $3.60/1M

Qwen3.5-35B-A3B

Novita

Input Price:Input: $0.25/1MOutput Price:Output: $2.00/1M

* Prices shown are per million tokens

Outputs Comparison

Notice missing or incorrect data?Start an Issue discussion→

Key Takeaways

Qwen3.5-397B-A17B

View details

Alibaba Cloud / Qwen Team

Higher AA-LCR score (68.7% vs 58.5%)

Higher BFCL-V4 score (72.9% vs 67.3%)

Higher BrowseComp score (69.0% vs 61.0%)

Higher BrowseComp-zh score (70.3% vs 69.5%)

Higher C-Eval score (93.0% vs 90.2%)

Higher DeepPlanning score (34.3% vs 22.8%)

Higher Global PIQA score (89.8% vs 86.6%)

Higher GPQA score (88.4% vs 84.2%)

Higher HMMT 2025 score (94.8% vs 89.0%)

Higher HMMT25 score (92.7% vs 89.2%)

Higher IFBench score (76.5% vs 70.2%)

Higher IFEval score (92.6% vs 91.9%)

Higher Include score (85.6% vs 79.7%)

Higher LiveCodeBench v6 score (83.6% vs 74.6%)

Higher LongBench v2 score (63.2% vs 59.0%)

Higher MAXIFE score (88.2% vs 86.6%)

Higher MMLU-Pro score (87.8% vs 85.3%)

Higher MMLU-ProX score (84.7% vs 81.0%)

Higher MMLU-Redux score (94.9% vs 93.3%)

Higher MMMLU score (88.5% vs 85.2%)

Higher Multi-Challenge score (67.6% vs 60.0%)

Higher NOVA-63 score (59.1% vs 57.1%)

Higher PolyMATH score (73.3% vs 64.4%)

Higher Seal-0 score (46.9% vs 41.4%)

Higher SuperGPQA score (70.4% vs 63.4%)

Higher SWE-Bench Verified score (76.4% vs 69.2%)

Higher t2-bench score (86.7% vs 81.2%)

Higher Terminal-Bench 2.0 score (52.5% vs 40.5%)

Higher VITA-Bench score (49.7% vs 31.9%)

Higher WideSearch score (74.0% vs 57.1%)

Higher WMT24++ score (78.9% vs 76.3%)

Qwen3.5-35B-A3B

View details

Alibaba Cloud / Qwen Team

Less expensive input tokens

Less expensive output tokens

Higher Humanity's Last Exam score (47.4% vs 28.7%)

Detailed Comparison

AI Model Comparison Table
Feature	Qwen3.5-397B-A17B	Qwen3.5-35B-A3B