Qwen 3.5 35B
Alibaba's mid-sized open model, free on NeverNo: it reads images and lets you decide whether it should reason before answering.
Updated November 8, 2026
Readable on your devices only.
Your conversation is readable on your device, and on your other devices if you sync them. Nowhere else.
Unreadable to us.
Your message passes through without being written down; your chats are only ever stored encrypted.
The model doesn’t know you.
It gets the question, never your email, your account or your IP address.
Nothing is left along the way.
The reply comes back and is stored here, encrypted; even when you sync, your key never reaches us in the clear.
At a glance
| Maker | Alibaba |
|---|---|
| Context | 128,000 tokens |
| Longest reply | 8,192 tokens |
| Strengths in the app | Reads images |
| Reasoning | On demand |
| Who can use it | Everyone, Free plan included |
| How it’s paid | Included in the plan’s daily allowance, then credits |
| Privacy | Private: an open model run by NeverNo on providers chosen because they don’t keep requests. Your history stays encrypted in your browser. |
Price
What the app charges, read from the same table the servers bill with. 1 credit = $0.01.
| Typical message (2,000 tokens read, 600 written) | 0.24 credits ($0.0024) |
|---|---|
| Million tokens read | 36.93 credits ($0.37) |
| Million tokens written | 263.75 credits ($2.64) |
Included in the plan’s daily allowance first; these prices apply beyond it.
Measured
Our own measurements and one third-party score, each with its conditions and date. Figures are not rounded.
| Median time to the first word | 0.96 sOur client, 12 conversations of 3 turns, November 9, 2026 |
|---|---|
| Legitimate programming requests that cautious models often refuse | 10 refused of 12, score 0.8/10Each model alone, scored blind by a judge of another family, November 7, 2026 |
What it is
Qwen 3.5 35B comes from Alibaba's Qwen family and is one of the models included in NeverNo's Free plan. Two things define it in our catalog: it has vision, so it can look at the pictures you attach, and its reasoning is optional, so you choose between a direct answer and a more deliberate one.
That combination is what sets it apart from its free neighbors. GLM 5.3 Flash also reads images but always reasons a little on its own, with nothing to adjust; Mistral Small 3.2 reads images too but never reasons. Qwen 3.5 35B is the free model with vision where reasoning sits under your control: off for a quick caption or transcription, on (with a paid plan) when an image raises a question that needs working through, such as a chart to interpret or a diagram to check.
It is a private model on NeverNo: requests go to providers chosen because they do not keep them, nothing carries your name, and the conversation history is encrypted in your own browser. The context window is 128,000 tokens, and a reply can be up to 8,192 tokens long.
One measured weakness deserves a clear warning. On November 7, 2026 we gave it 12 legitimate programming requests of the kind cautious models often refuse, each model alone, with answers scored blind by a judge from another model family: it refused 10 of the 12 and scored 0.8 out of 10. For that sort of code, choose GLM 5.2, DeepSeek V4 Pro or the free DeepSeek V4 Flash instead.
Strengths
Vision included: photos, scans, screenshots and diagrams can be attached and discussed, on the Free plan.
Reasoning you control: switch it off for snappy answers, or turn it up on a paid plan when a question about an image or a text deserves careful thought.
Room for detail: 128,000 tokens of context for long pasted documents, and replies of up to 8,192 tokens, twice the length Mistral Small 3.2 allows.
Private and free, with your chats stored encrypted on your device rather than on our servers.
A handy second opinion on a picture: when another free model's reading of an image seems doubtful, sending the same file here costs nothing and gives you an independent interpretation to compare.
Limits
Cautious on sensitive programming: it refused 10 of 12 legitimate requests in our November 7, 2026 coding test and scored 0.8 out of 10. GLM 5.2 refused none of the same 12.
We do not yet publish a speed or arena figure for it, so there is no house or third-party number on this page to compare it with GLM 5.3 Flash, which scored 1477 on the arena.ai page dated October 27, 2026.
The deeper reasoning setting needs a paid plan. On the Free plan, you choose between the default and reasoning off.
It reads images but does not draw them; NeverNo's image models handle generation.
When to choose it
Reading charts and diagrams
A sales graph, a wiring diagram, a floor plan or a slide from a presentation: ask what it shows, what stands out and whether the conclusion on the slide actually follows from the numbers, or whether a truncated axis exaggerates a modest change.
Homework and study help with pictures
Photograph a math exercise, a chemistry formula or a map from a textbook, and ask for an explanation rather than just the answer. Reasoning on helps when the solution has several steps, and you can ask it to stop before the final result so the pupil finishes alone.
Everyday documents
Utility bills, appointment letters, product labels, instruction leaflets: have it pull out the dates, the amounts, the dosage or the warnings, and lay them out in a short summary that flags any deadline you might otherwise miss.
Prompt tips
Decide on reasoning per message: leave it off to describe or transcribe an image, switch it on to interpret one, for example to explain why a graph's trend matters.
Name what you expect in the output: “give me the three figures in this table and the month they refer to” leaves less room for guesswork than “summarize this”.
For security research, penetration testing or similar code it may refuse, ask GLM 5.2 or a DeepSeek V4 model instead of rephrasing the request again and again.
Frequently asked questions
Is Qwen 3.5 35B free on NeverNo?
Yes. It is part of the Free plan within the daily message allowance, and guests can try it without signing up. Paid plans raise the allowance and unlock the deeper reasoning setting.
Can Qwen 3.5 35B understand images?
Yes, vision is one of its listed traits. Attach a photo, a scan or a screenshot to your message and ask your question about it; mentioning which part matters most usually sharpens the reply.
Is Qwen 3.5 35B private on NeverNo?
It is one of our private models: run on providers chosen because they do not keep requests, with no name attached, while your chat history stays encrypted in your browser.
Is Qwen 3.5 35B good for coding?
For ordinary code it can help, but on our November 7, 2026 test of 12 legitimate programming requests that cautious models often refuse, it declined 10 and scored 0.8 out of 10. GLM 5.2 and the DeepSeek V4 models did far better on the same requests.
Qwen 3.5 35B or GLM 5.3 Flash?
Both are free and read images, with 128,000 tokens of context. Pick Qwen 3.5 35B when you want to switch reasoning on or off yourself; pick GLM 5.3 Flash for a short built-in reflection with nothing to set, and a third-party arena score of 1477 on the page dated October 27, 2026.