Qwen: Qwen3.8 2.4T A95B.
Qwen3.8-2.4T-A95B is a text mixture-of-experts model, 2.4T total and 95B active, with open weights under Qwen's own licence, aimed at long agentic work.
8× AMD MI325 · $25.36/hr.
Most-aggressive quantisation we have a working recommendation for. Lower precision = less VRAM = cheaper hardware, at a small accuracy cost.
Cheapest hosted endpoints.
Variants in the Qwen family.
Qwen3.8-Flash
Qwen3.8-Flash is Qwen's official API model built on Qwen3.8-Flash-Next, with ...
Qwen 2.5 72B
Alibaba's flagship open-weight LLM — 72B dense.
Qwen 2.5 7B
Qwen2.5-7B-Instruct is a 7.61B instruction-tuned open-weights text model unde...
Qwen 2.5 14B
14B Qwen 2.5 — sweet spot for single-GPU local hosting.
Qwen 2.5 32B
32B Qwen 2.5 — laptop-class workhorse.
Qwen 2.5 3B
3B Qwen 2.5 — laptop / edge target.
Qwen: Qwen3.8 27B
Qwen3.8-27B is a 27B dense Apache 2.0 model with image and video input and a ...
Qwen 3 235B
Alibaba's frontier MoE — 235B total / 22B active.
Qwen 3 32B
Qwen3-32B is a dense 32.8B open-weights text model under Apache 2.0 that swit...
Qwen 3 14B
Qwen3-14B is a dense 14.8B open-weights text model under Apache 2.0 that swit...
Qwen 3 8B
Qwen3-8B is a dense 8.2B parameter causal language model from the Qwen3 serie...
Qwen 3 4B
Qwen: Qwen3.8 Max
Qwen3.8 Max is the flagship model in Alibaba's Qwen3.8 series, the general-av...
Qwen: Qwen3.7 Flash
Qwen3.7 Flash is a vision-language reasoning model from Alibaba. It is suited...
Qwen: Qwen3.7 Plus
Qwen3.7-Plus is an Alibaba model with image, video and text input and a 1M-to...
Qwen: Qwen3.8 Max (0902)
Qwen3.8 Max 0902 is an updated snapshot of Qwen3.8 Max from Alibaba's Qwen te...
Qwen 2 (1.5B)
Qwen 2 (72B)
Qwen 2 (7B)
Qwen 2 Instruct (1.5B)
Qwen2.5 1.5B
Qwen2.5 1.5B Instruct
Qwen2.5 14B
Qwen2.5 32B
Qwen2.5 72B
Qwen2.5 7B
Qwen3 0.6B
Qwen3 0.6B Base
Qwen3 1.7B
Qwen3 1.7B Base
Qwen3 14B Base
Qwen3 30B A3b Base
Qwen3 4B Base
Qwen3 4B Instruct 2507
Qwen3 8B Base
Qwen3.5-0.8B
Qwen3.5-0.8B is Alibaba's smallest model in the Qwen3.5 series, featuring a h...
Qwen3.5-2B
Qwen3.5-2B is a compact yet capable model from Alibaba's Qwen3.5 series. It f...
Qwen3.5-4B
Qwen3.5-4B is a mid-size model from Alibaba's Qwen3.5 series that delivers a ...
Qwen: Qwen Plus 0728
Qwen Plus 0728, based on the Qwen3 foundation model, is a 1 million context h...
Qwen: Qwen-Plus
Qwen-Plus is the Plus model of Alibaba's Qwen3 series, text only with a 1M-to...
Qwen: Qwen3 235B A22B Instruct 2507
Qwen3-235B-A22B-Instruct-2507 is a multilingual, instruction-tuned mixture-of...
Qwen: Qwen3 235B A22B Thinking 2507
Qwen3-235B-A22B-Thinking-2507 is a high-performance, open-weight Mixture-of-E...
Qwen: Qwen3 30B A3B
Qwen3-30B-A3B is an Apache 2.0 text model with 30.5B total and 3.3B active pa...
Qwen: Qwen3 30B A3B Instruct 2507
Qwen3-30B-A3B-Instruct-2507 is a 30.5B-parameter mixture-of-experts language ...
Qwen: Qwen3 30B A3B Thinking 2507
Qwen3-30B-A3B-Thinking-2507 is a 30B parameter Mixture-of-Experts reasoning m...
Qwen: Qwen3 Max
Qwen3-Max is an updated release built on the Qwen3 series, offering major imp...
Qwen: Qwen3 Next 80B A3B Instruct
Qwen3-Next-80B-A3B-Instruct is an instruction-tuned chat model in the Qwen3-N...
Qwen: Qwen3 Next 80B A3B Thinking
Qwen3-Next-80B-A3B-Thinking is a reasoning-first chat model in the Qwen3-Next...
Qwen: Qwen3.5 397B A17B
The Qwen3.5 series 397B-A17B native vision-language model is built on a hybri...
Qwen: Qwen3.5 Plus 2026-02-15
The Qwen3.5 native vision-language series Plus models are built on a hybrid a...
Qwen: Qwen3.5 Plus 2026-04-20
Qwen3.5-Plus 2026-04-20 is a dated snapshot with image, video and text input ...
Qwen: Qwen3.5-122B-A10B
The Qwen3.5 122B-A10B native vision-language model is built on a hybrid archi...
Qwen: Qwen3.5-27B
The Qwen3.5 27B native vision-language Dense model incorporates a linear atte...
Qwen: Qwen3.5-35B-A3B
The Qwen3.5 Series 35B-A3B is a native vision-language model designed with a ...
Qwen: Qwen3.5-9B
Qwen3.5-9B is a multimodal foundation model from the Qwen3.5 family, designed...
Qwen: Qwen3.5-Flash
The Qwen3.5 native vision-language Flash models are built on a hybrid archite...
Qwen: Qwen3.6 27B
Qwen3.6-27B is a 27B dense open-weights model, Apache 2.0, with image and tex...
Qwen: Qwen3.6 35B A3B
Qwen3.6-35B-A3B is an Apache 2.0 mixture-of-experts model, 35B total and 3B a...
Qwen: Qwen3.6 Flash
Qwen3.6-Flash is a vision-language model on Qwen's API that takes text, image...
Qwen: Qwen3.6 Plus
Qwen 3.6 Plus builds on a hybrid architecture that combines efficient linear ...
Qwen: Qwen3.7 Max
Qwen3.7-Max is Alibaba's largest Qwen3.7 model, a text-only model with no dow...
Qwen: Qwen3.6 Max Preview
Qwen3.6-Max-Preview is a proprietary frontier model from Alibaba Cloud built ...
Cheapest hardware per quantisation.
Each row is one quantisation tier (the same weights compressed differently). Lower precision → lower VRAM → cheaper hardware, at the cost of small accuracy loss. $/hr refreshed hourly from each provider's API.
| Quantisation | Cheapest GPU config | Total VRAM | Live $/hr | tokens/sec | |
|---|---|---|---|---|---|
|
INT4
INT4 — 4-bit integer (~4× VRAM saving)
|
2048 GB | $25.36/hr | — | Compare → |
What it costs per month across providers.
Estimate your monthly bill for Qwen: Qwen3.8 2.4T A95B across every host that publishes per-token pricing. Slide your token volumes; the chart + table re-rank cheapest-first.
Cheapest provider on the left.
Total monthly cost — input + output tokens combined.
Bill breakdown.
Rent the GPU instead of paying per token.
For an open-weights model like Qwen: Qwen3.8 2.4T A95B, you can rent a GPU and serve inference yourself. The math: cheapest GPU rental × 730 hours/month + your electricity rate × power draw.
Assumes the GPU runs 24/7 at ~85% utilisation. If your traffic is bursty, you'll pay less for the API and probably more for the GPU (idle hours still cost rental). The breakeven analysis lives on the Self-host vs API breakeven tool.
About Qwen: Qwen3.8 2.4T A95B.
Qwen3.8-2.4T-A95B is the post-trained open-weights model in Alibaba's Qwen3.8 generation, under Qwen's own licence. It is a mixture-of-experts text model with 2.4T parameters in total and 95B active. Context is 262,144 tokens natively, extendable to about 1 million. Qwen aims it at coding, professional work, research and long agentic tasks. Reasoning depth can be tuned, and earlier reasoning can be kept across turns. Vision input and a non-thinking mode are only in the hosted Qwen3.8-Max.