Qwen3.5-35B-A3B
AlibabaReleased Feb 23, 2026Open weights · Apache 2.0qwen3.5-35b-a3b
- Intelligence (Epoch)
- 142 · 73rd of 111
- Reasoning
- yes
- Knowledge cutoff
- September 2025
- Context
- 262K tokens
- Max output
- 82k tokens
- Input
- text, image, video
- API features
- tools, structured outputs, JSON mode, seed, logprobs
- Alibaba's API
- yes, as qwen3.5-35b-a3b
Read Oct 23, 06:00 UTC from the hosts' published offers and OpenRouter, spec sheet from Alibaba · cutoff: aiknowledgecutoff.com
The cheapest host, DeepInfra, serves fp8 weights (maker: bf16). Weights and context per host
Intelligence for the price
13 models score higher for the same price or less; it ranks 14 of the 37 models at its price or below.
Epoch Capabilities Index against the lowest input + output price per M, log scale · prices Oct 23. Score: Epoch AI, CC BY 4.0, 142 with an interval of 140 to 145.
Against its rivals
The best of its price range and the top score
- Qwen3.5-35B-A3Bthis model$1.14142
- Claude Opus 5.5top score$24.00169
- Claude Haiku 5.5same price range$0.60159
- GPT-6 Lunasame price range$0.60157
Same level for 7.1× less: Qwen3.7 Flash at Alibaba, $0.16 per M input + output · their Epoch intervals overlap (144 against 142).
Bars: Epoch Capabilities Index (Epoch AI, CC BY 4.0), drawn from 100, below the lowest score in the catalogue · price: lowest input + output per M.
Speed by provider
Output speed of Qwen3.5-35B-A3B at each host, fastest first
- Alibaba127 tok/s0.50 s to first token
- Venice111 tok/s0.88 s to first token
- Parasail98 tok/s1.00 s to first token
- SiliconFlow43 tok/s1.60 s to first token
- AtlasCloud40 tok/s2.06 s to first token
- DeepInfra26 tok/s0.65 s to first token
Speed: OpenRouter, median of the last 30 min, read Oct 23, 06:00 UTC · a snapshot of real traffic before the daily read, not a daily average · bars from zero.
Run it yourself
open weights, Apache 2.0 · fits 6 of the 6 workstations in our guideSelf-host guide ›- Parameters
- 35B
- 3B active per token (MoE)
- Memory needed
- 21 GB
- in INT4 · 42 GB in INT8
- Smallest cloud setup
- RTX 3090
- 24 GB · $0.17/h, about $124 a month
- Worth it above
- 109M tokens
- a month, against the cheapest API at $1.14 per M
| Quantization | Memory | Smallest cloud GPU | Quality |
|---|---|---|---|
| INT4 (Q4) | 21 GB | RTX 3090 · $0.17/h | smallest footprint, minor quality loss |
| INT8 (Q8) | 42 GB | A100 · $1.19/h | near-lossless |
| FP16 (full) | 84 GB | 2× H100 · $4.30/h | reference quality |
At home: in INT4 it fits a RTX 5090 (32 GB) + host PC (plus a host PC), about $3,500 once, or $0.1332/h over 3 years, electricity aside; it pays for itself after about 3B tokens against the cheapest API.
Memory ≈ parameters × bytes per parameter × 1.2 · cloud price: on-demand rate, running 24/7 · builds: GGUF, AWQ, MLX, GPTQ, Ollama · break-even calculator
| Provider | Input | Output | Cache ↑ | tok/s | First token | Uptime 24h | Checked |
|---|---|---|---|---|---|---|---|
| DeepInfrafp8 (bf16) · 256K ctx · 80K out | $0.14 | $1.00 | $0.05 | 26 | 0.65 s | 99.9% | Oct 23 |
| Parasailfp8 (bf16) · 256K ctx | $0.15 | $1.00 | $0.05 | 98 | 1.00 s | 100% | Oct 23 |
| Veniceprecision not read: host terms · 250K ctx · 16K out | $0.15 | $1.00 | $0.05 | 111 | 0.88 s | 100% | Oct 23 |
| SiliconFlowfp8 (bf16) · 256K ctx | $0.24 | $1.80 | $0.15 | 43 | 1.60 s | 85.7% | Oct 23 |
| AtlasCloudfp8 (bf16) · 256K ctx · 64K out | $0.23 | $1.80 | $0.23 | 40 | 2.06 s | 100% | Oct 23 |
| CoreWeave256K ctx | $0.25 | $1.25 | $0.25 | — | — | — | Sep 3 |
| Alibabaprecision not read: host terms · 256K ctx · 64K out | $0.25 | $2.00 | — | 127 | 0.50 s | 99.8% | Oct 23 |
| Novita AIprecision not stated · 256K ctx · 64K out | $0.25 | $2.00 | — | — | — | — | Oct 23 |
8 of 8 providers · prices per million tokens, read Oct 23, 06:00 UTC · speed: OpenRouter, median of the last 30 min, read Oct 23, 06:00 UTC · uptime: OpenRouter, last 24 h · under each host: weights precision, context and max output as served, the maker's value in brackets when the host's differs
| Provider | Weights | Context | Max output | Tools | Structured outputs | Reasoning |
|---|---|---|---|---|---|---|
| DeepInfra | fp8 | 256K | 80K | no | yes | yes |
| Parasail | fp8 | 256K | not read: host terms | yes | yes | yes |
| Venice | not read: host terms | 250K | 16K | yes | yes | yes |
| CoreWeave | not checked yet | 256K | not checked yet | — | — | — |
| AtlasCloud | fp8 | 256K | 64K | yes | yes | yes |
| SiliconFlow | fp8 | 256K | not read: host terms | no | yes | yes |
| Alibaba | not read: host terms | 256K | 64K | yes | yes | yes |
| Novita AI | not stated by the host | 256K | 64K | — | — | — |
Weights, context and max output: read at each host's own catalogue or docs where its terms allow, otherwise as the host declares to OpenRouter; hover a cell for its source and date · tools, structured outputs, reasoning: as declared to OpenRouter · not verified by us
Qwen3.5-35B-A3B spec sheet, field by field, with sources
Qwen3.5-35B-A3B spec sheetmaker-declared
What Alibaba declares about the model itself — read from its own weights repository and documentation, not from any host. What each endpoint actually serves is in the pricing above: the two can differ, and both are true.
- Context window
- 262,144 tokens
- Parameters
- 35,951,822,704≈ 36.0B
- Active experts
- 8 of 256 active per token
- Weight precision
- BF16
- License
- Apache 2.0 — open weights, permissivecurated · read 2026-09-07Scope: the licence on the published weights, and nothing else. It is not the maker's acceptable-use policy, which is a separate document, and it is not the contract you sign with whichever host you call — hosts set their own terms, which this index does not read.
- Knowledge cutoff
- not published by the maker≈ September 2025 third-party estimate — not the maker's figurecurated
- Modalities
- text · image · video
- Max output (official)
- to verify
- Release date
- to verify≈ February 23, 2026 third-party catalogue — not the maker's figure
- Maker lifecycle status
- to verify
Also declared in the weights
- Architecture class
- Qwen3_5MoeForConditionalGeneration
- Layers
- 40
- Vocabulary
- 248,320 tokens
- Weights first published
- February 24, 2026
- Weights last updated
- April 24, 2026
Qwen/Qwen3.5-35B-A3B · config.json @ 59d61f3 · read 2026-10-08 · 6 of 10 fields published by the maker
| Date | Provider | Input | Output | Change |
|---|---|---|---|---|
| Oct 9 | Venice | $0.31 → $0.15 | $1.25 → $1.00 | −26% |
Every published change of input or output price, newest first · change on input + output · full series: pricing-history.csv.
| Provider | Chatbot 1,000 in / 500 out | RAG / long context 8,000 in / 500 out | Batch summarize 4,000 in / 1,000 out |
|---|---|---|---|
| DeepInfra | $0.64 | $1.62 | $1.56 |
| Parasail | $0.65 | $1.70 | $1.60 |
| Venice | $0.65 | $1.70 | $1.60 |
| CoreWeave | $0.88 | $2.62 | $2.25 |
| AtlasCloud | $1.12 | $2.70 | $2.70 |
| SiliconFlow | $1.14 | $2.82 | $2.76 |
| Alibaba | $1.25 | $3.00 | $3.00 |
| Novita AI | $1.25 | $3.00 | $3.00 |
What each host bills for 1,000 requests of a representative job
Your own workload
Other Alibaba models · all its models, hosts and history ›
- Qwen3.8 27B16 providers · from $1.29
- Qwen3.6 35B A3B12 providers · from $1.00
- Qwen3 235B A22B Instruct 250710 providers · from $0.44
- Qwen3.5 397B A17B10 providers · from $3.45
- Qwen3.6 27B9 providers · from $2.25
- Qwen3 30B A3B7 providers · from $0.24