Qwen3.8 2.4T A95B
Qwen · released Aug 8, 2026 · Qwen/Qwen3.8-2.4T-A95B
- Type
- Open weightsCustom licence
- Params
- 2.4T
- Context
- 262K
95B active per word · about 197K words of context · download allowed, licence restricts use
Our take
Written Sep 17, 2026Qwen3.8 2.4T A95B is a text-in, text-out model you can download and run yourself, and ten hosts serve it at one shared rate. No licence is supplied and no benchmark scores are, so the terms and the quality both need checking before you commit.
Use it for long-document work where a whole report or a stack of files goes into one request instead of being chunked, and where you are willing to trial it on your own tasks. Because every listed host charges the same rate, choosing between them comes down to speed and reliability rather than price. Skip it if you need a licence you can read before building a commercial product on it, or if you need measured evidence of quality.
The case for it
- You can download it and run it yourself, so a host is optional and you are not tied to one company's uptime or terms.
- The request capacity takes a long report or a stack of documents beside the question, though reliable recall across all of it is unverified in our data.
- All ten listed hosts charge the same rate, so picking between them is a question of speed and reliability, neither of which is measured here.
The case against it
- No licence is supplied, so whether commercial use, changes and redistribution are permitted is unverified in our data and would need checking at the source.
- No benchmark scores are supplied, so there is no evidence of coding, reasoning or chat ability and the only way to judge it is a trial on work you can check yourself.
- The parameter count is not published, so nothing here indicates what running it yourself would require.
How good is it?
We hold no score for this model.
So there is no figure here for everyday use, coding, agent work or writing. That is a gap in our data, not a low score.
Can you run it yourself?
GeForce RTX 4090 · 24 GB
Too large for this card. The weights do not fit even with part of them offloaded to system memory.
Apple M1 Pro (16-core GPU) · 32 GB
Too large for this card. The weights do not fit even with part of them offloaded to system memory.
Radeon RX 7900 XT · 20 GB
Too large for this card. The weights do not fit even with part of them offloaded to system memory.
Memory use by level
Against a 24 GB card.
What is quantisation? →This model on every device we track71 devicesThe Q4 build most people download, on each device: what the weights come to, how much context the memory leaves, and whether it runs. Smallest device that runs it first.
Check against your own machine → · Where to rent it hosted →
Or rent it from someone else
Prices checked 4 hours ago — each listing carries its own date.
- per 1M tokens
- $2.00 in / $6.00 out
- Context served
- 1M
- Throughput
- Not measured
| Provider | In / out per 1M tokens | Context | Throughput | Trains on prompts | Logs prompts | Zero retention |
|---|---|---|---|---|---|---|
| OpenRouterOpenRouter's own listing | $2.00 / $6.00checked 4 hours ago | 1M | not measured | Unknown | Unknown | Unknown |
| DeepInfrafp4Direct and through OpenRouter | $2.00 / $6.00checked 4 hours ago | 262K131K max reply through OpenRouter | 82 tok/sthrough OpenRouter | DirectUnknownThrough OpenRouterNo | DirectUnknownThrough OpenRouterNo | DirectUnknownThrough OpenRouterConfirmed |
| Novita AIDirect and through OpenRouter | $2.00 / $6.00checked 4 hours ago | 1M131K max reply through OpenRouter | 33 tok/sthrough OpenRouter | DirectUnknownThrough OpenRouterNo | DirectUnknownThrough OpenRouterNo | DirectUnknownThrough OpenRouterConfirmed |
| Together AIThrough OpenRouter | $2.00 / $6.00checked 4 hours ago | 1M909K max reply | 146 tok/s | No | No | Confirmed |
| SiliconFlowfp8Through OpenRouter | $2.00 / $6.00checked 4 hours ago | 1M131K max reply | 31 tok/s | No | No | Confirmed |
| Venice AIThrough OpenRouter | $2.00 / $6.00checked 4 hours ago | 262K66K max reply | 54 tok/s | No | No | Confirmed |
| Alibaba CloudThrough OpenRouter | $2.00 / $6.00checked 4 hours ago | 1M131K max reply | 42 tok/s | No | Yesunknown period | Unknown |
| ModalThrough OpenRouter | $2.00 / $6.00checked 4 hours ago | 1M262K max reply | 113 tok/s | No | No | Confirmed |
DeepInfra
Novita AI
Together AIAcross the 8 listings we hold: 7 say they do not train on prompts (2 of them only through OpenRouter), 0 say they do and 1 does not say. 6 appear in the zero-retention registry we check (2 of them only through OpenRouter); the rest are unknown to us.
What each host's API supports
From the parameter list each endpoint publishes. Streaming is omitted: nothing we hold reports it, for any model.
| Provider | Tool calling | JSON output | Strict schema |
|---|---|---|---|
| OpenRouterOpenRouter's own listing | ✓ | ✓ | ✓ |
| DeepInfrafp4Direct and through OpenRouter | ✓ | ✓ | ✓ |
| Novita AIDirect and through OpenRouter | ✓ | ✓ | ✓ |
| Together AIThrough OpenRouter | ✓ | ✓ | ✓ |
| SiliconFlowfp8Through OpenRouter | ✓ | ✓ | ✓ |
| Venice AIThrough OpenRouter | ✓ | ✗ | ✗ |
| Alibaba CloudThrough OpenRouter | ✓ | ✓ | ✓ |
| ModalThrough OpenRouter | ✓ | ✓ | ✓ |
Tool calling: 8 of 8 listings say yes. JSON output: 7 of 8 listings say yes, 1 says no. Strict schema: 7 of 8 listings say yes, 1 says no.
Models people weigh against Qwen3.8 2.4T A95B
When we formed this view
Recent changes
What moved
input +11% ($1.80 → $2.00 per 1M tokens), output +11% ($5.40 → $6.00 per 1M tokens), cache read +11% ($0.18 → $0.20 per 1M tokens)What moved
first indexed by our pipelineEach date is the day we first saw the change, or the day the maker announced it.
What we do not know about this model yet
- We hold no measured file for it, so all 3 sizes on this page are calculated from the parameter count.
- No independent board has scored it, so we hold no quality figures at all.
- Nothing we hold says whether an endpoint streams, so we do not show it either way.
- 1 of 8 listings does not say whether it trains on prompts, and 2 answer only through OpenRouter, not for their own listing.
- We hold no batch or off-peak rate for any of its listings.
- We hold a decode speed for it, but no prompt-processing (prefill) figure, so how long the input side of a job takes is unknown to us.
Licence and identifiers
What the licence allowsCustom licence, what it allows commercially, and the identifiers you need to pull this model — its Hugging Face repo, our slug and a machine-readable card.
Licence
Custom licence
This model ships custom license terms that don't map to a known template. We haven't parsed them, so commercial use, redistribution and derivatives are unverified — review the original terms before shipping.
Identifiers
- Hugging Face
- Qwen/Qwen3.8-2.4T-A95B
- Architecture
- Mixture of experts
- Takes in, gives back
- Text in, text out
- Catalogue slug
- qwen-qwen3-8-2-4t-a95b