Qwen3.6 35B A3B
AlibabaReleased Apr 17, 2026Open weights · Apache 2.0qwen3.6-35b-a3b
- Intelligence (Epoch)
- 144 · 68th of 111
- Reasoning
- yes
- Knowledge cutoff
- September 2025
- Context
- 262K tokens
- Max output
- 128k tokens
- Input
- text, image, video
- API features
- tools, structured outputs, JSON mode, seed, logprobs
Read Oct 23, 06:00 UTC from the hosts' published offers and OpenRouter, spec sheet from Alibaba · cutoff: aiknowledgecutoff.com
The cheapest host, AkashML, serves fp8 weights (maker: bf16). Weights and context per host
Intelligence for the price
8 models score higher for the same price or less; it ranks 9 of the 35 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, 144 with an interval of 141 to 146.
Against its rivals
Its previous version, the best of its price range and the top score
- Qwen3.6 35B A3Bthis model$1.00144
- Claude Opus 5.5top score$24.00169
- Claude Haiku 5.5same price range$0.60159
- GPT-6 Lunasame price range$0.60157
- Qwen3.5-35B-A3Bprevious version$1.14142
Same level for 6.3× less: Qwen3.7 Flash at Alibaba, $0.16 per M input + output · their Epoch intervals overlap (144 against 144).
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 Qwen3.6 35B A3B at each host, fastest first
- DekaLLM176 tok/s0.49 s to first token
- AtlasCloud156 tok/s0.44 s to first token
- Venice116 tok/s0.72 s to first token
- CoreWeave113 tok/s0.33 s to first token
- SiliconFlow94 tok/s1.04 s to first token
- Phala91 tok/s0.75 s to first token
- Parasail81 tok/s0.70 s to first token
- AkashML66 tok/s0.55 s to first token
- DeepInfra33 tok/s1.19 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
- 124M tokens
- a month, against the cheapest API at $1.00 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 4B 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 |
|---|---|---|---|---|---|---|---|
| AkashMLfp8 (bf16) · 256K ctx | $0.10 | $0.90 | $0.05 | 66 | 0.55 s | 99.8% | Oct 23 |
| Parasailfp8 (bf16) · 256K ctx | $0.10 | $0.95 | $0.05 | 81 | 0.70 s | 100% | Oct 23 |
| DekaLLM256K ctx | $0.10 | $1.00 | $0.05 | 176 | 0.49 s | 99.6% | Oct 23 |
| Rekaprecision not read: host terms · 256K ctx · 128K out | $0.10 | $1.00 | $0.05 | — | — | 99.6% | Oct 1 |
| Phalaprecision not read: host terms · 256K ctx | $0.20 | $1.27 | $0.056 | 91 | 0.75 s | 98.5% | Oct 23 |
| DeepInfrafp8 (bf16) · 256K ctx · 64K out | $0.10 | $0.95 | $0.10 | 33 | 1.19 s | 99.1% | Oct 23 |
| Io Netprecision not read: host terms · 256K ctx | $0.22 | $1.20 | $0.12 | — | — | 87.7% | Sep 22 |
| SiliconFlowfp8 (bf16) · 256K ctx | $0.24 | $1.80 | $0.15 | 94 | 1.04 s | 99.4% | Oct 23 |
| AtlasCloudfp8 (bf16) · 256K ctx · 64K out | $0.25 | $1.25 | $0.25 | 156 | 0.44 s | 70.2% | Oct 23 |
| CoreWeavefp8 (bf16) · 256K ctx | $0.25 | $1.25 | $0.25 | 113 | 0.33 s | 100% | Oct 23 |
| Venicefp8 (bf16) · 250K ctx · 64K out | $0.10 | $1.00 | — | 116 | 0.72 s | 99.9% | Oct 23 |
| Novita AIprecision not stated · 256K ctx · 64K out | $0.25 | $1.49 | — | — | — | — | Oct 23 |
10 of 12 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 |
|---|---|---|---|---|---|---|
| AkashML | fp8 | 256K | not stated by the host | yes | yes | yes |
| DeepInfra | fp8 | 256K | 64K | yes | yes | yes |
| Parasail | fp8 | 256K | not read: host terms | yes | yes | yes |
| Venice | fp8 | 250K | 64K | yes | no | yes |
| DekaLLM | not checked yet | 256K | not checked yet | yes | yes | yes |
| Reka | not read: host terms | 256K | 128K | — | — | — |
| Io Net | not read: host terms | 256K | not read: host terms | — | — | — |
| Phala | not read: host terms | 256K | not read: host terms | yes | yes | yes |
| AtlasCloud | fp8 | 256K | 64K | yes | no | yes |
| CoreWeave | fp8 | 256K | not checked yet | no | yes | yes |
| Novita AI | not stated by the host | 256K | 64K | — | — | — |
| SiliconFlow | fp8 | 256K | not read: host terms | no | yes | yes |
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
Qwen3.6 35B A3B spec sheet, field by field, with sources
Qwen3.6 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≈ April 17, 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
- April 15, 2026
- Weights last updated
- April 24, 2026
Qwen/Qwen3.6-35B-A3B · config.json @ 995ad96 · read 2026-10-08 · 6 of 10 fields published by the maker
| Date | Provider | Input | Output | Change |
|---|---|---|---|---|
| Oct 20 | Parasail | $0.15 → $0.10 | $1.00 → $0.95 | −9% |
| Oct 17 | AtlasCloud | $0.19 → $0.25 | $1.11 → $1.25 | +15% |
| Sep 22 | Io Net | $0.19 → $0.22 | $0.99 → $1.20 | +20% |
| Sep 16 | Io Net | $0.13 → $0.19 | $0.94 → $0.99 | +10% |
| Sep 10 | SiliconFlow | $0.20 → $0.24 | $1.60 → $1.80 | +13% |
| Sep 9 | Io Net | $0.19 → $0.13 | $1.19 → $0.94 | −22% |
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 |
|---|---|---|---|
| AkashML | $0.55 | $1.25 | $1.30 |
| DeepInfra | $0.57 | $1.28 | $1.35 |
| Parasail | $0.57 | $1.28 | $1.35 |
| Venice | $0.60 | $1.30 | $1.40 |
| DekaLLM | $0.60 | $1.30 | $1.40 |
| Reka | $0.60 | $1.30 | $1.40 |
| Io Net | $0.82 | $2.36 | $2.08 |
| Phala | $0.84 | $2.23 | $2.07 |
| AtlasCloud | $0.88 | $2.62 | $2.25 |
| CoreWeave | $0.88 | $2.62 | $2.25 |
| Novita AI | $0.99 | $2.73 | $2.48 |
| SiliconFlow | $1.14 | $2.82 | $2.76 |
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 235B A22B Instruct 250710 providers · from $0.44
- Qwen3.5 397B A17B10 providers · from $3.45
- Qwen3.6 27B9 providers · from $2.25
- Qwen3.5-35B-A3B8 providers · from $1.14
- Qwen3 30B A3B7 providers · from $0.24