Qwen3.5-122B-A10B: The Qwen3.5 Architecture at a Servable Size

What Qwen3.5-122B-A10B is

Qwen3.5-122B-A10B is the mid-size member of Alibaba's Qwen3.5 generation: 122 billion total parameters with 10 billion active per token, Apache 2.0 and self-hostable, with a 256K token context window. It shares the generation's architecture with the much larger Qwen3.5-397B-A17B — the same design decisions at roughly a third of the total size and a little over half the active parameters.

What it inherits from the generation

Alibaba describe the Qwen3.5 line as built on a unified vision-language foundation, trained with early fusion on multimodal tokens, which they say reaches parity with Qwen3 while outperforming the separate Qwen3-VL models across reasoning, coding, agents and visual understanding. The architecture pairs Gated Delta Networks with sparse mixture-of-experts — the mechanism behind the low activation ratio — and reinforcement learning was scaled across what Alibaba call million-agent environments with progressively harder task distributions. Language coverage runs to 201 languages and dialects.

Alibaba’s figures

Alibaba's mid-size comparison table covers GPT-5-mini, GPT-OSS-120B, Qwen3-235B-A22B and the three Qwen3.5 mid-size models. Reading the rows the card labels: LongBench v2 gives Qwen3.5-122B-A10B 60.2, Qwen3.5-27B 60.6 and Qwen3.5-35B-A3B 59.0, against GPT-5-mini's 56.8 and GPT-OSS-120B's 48.2; HMMT February 2025 gives 91.4, 92.0 and 89.0 against 89.2 and 90.0; HMMT November 2025 gives 90.3, 89.8 and 89.2 against 84.2 and 90.0; and LiveCodeBench v6 gives 78.9, 80.7 and 74.6 against 80.5 and 82.7. Qwen3-235B-A22B records 54.8, 85.1, 89.5 and 75.1 on the same four. These are Alibaba's own figures.

LMArena

On LMArena's text leaderboard Qwen3.5 122B A10b has 1416 Elo (rank 139); by category, 1458 Elo in Coding (rank 145), 1426 Elo in Math (rank 120), 1442 Elo in Expert (rank 134) and 1366 Elo in Creative Writing (rank 161). Qwen3.5-397B-A17B has 1441 Elo (rank 94).

Price and hosted variant

Around $0.40 per million input tokens through third-party providers, an aggregated figure rather than first-party. The hosted Qwen3.5-Plus corresponds to the 397B weights, not these, and has a 1M context and built-in tooling.

Architecture

The card lists 48 layers with a hidden dimension of 3,072, arranged as 12 repeats of three Gated DeltaNet layers followed by one gated-attention layer, each feeding a mixture-of-experts block. There are 256 experts, with 8 routed and 1 shared expert active per token and an expert intermediate dimension of 1,024. The linear-attention layers use 64 value heads and 16 query-key heads of dimension 128; the gated-attention layers use 32 query heads and 2 key-value heads of dimension 256. The repository holds the post-trained model in Hugging Face Transformers format, compatible with Transformers, vLLM, SGLang and KTransformers. The context is 262,144 tokens natively, extensible to 1,010,000. The token embedding is padded to 248,320 entries. Alibaba also describe near-100% multimodal training efficiency compared with text-only training, and asynchronous reinforcement-learning frameworks for large agent scaffolds.

The Qwen3.5 family

The Qwen3.5 open-weight releases covered on this site are the 397B-A17B flagship, the 122B-A10B and 35B-A3B mixture-of-experts models, and the dense 27B. Alibaba name hosted versions for two of them on Alibaba Cloud Model Studio: Qwen3.5-Plus for the 397B-A17B and Qwen3.5-Flash for the 35B-A3B. The models can also be tried in Qwen Chat at chat.qwen.ai.

Official Resources