// Alibaba (Qwen)
Qwen3.8-Max
A Alibaba (Qwen) model. 2 prices across 2 channels side by side, each with its source and check date – plus specs, licence, speed and benchmarks.
- Official in/out per 1M
- ¥12 / ¥36
- Lowest (official)
- ¥12 / ¥36
- Context
- 1M
- AA index
- 45.4
Prices checked 2026-10-10Official sources linked on every rowNo estimates
01Specs at a glance
Official values first; gaps filled from OpenRouter’s API where the vendor page is silent.
| Vendor | Alibaba (Qwen) |
|---|---|
| Context window | 1M (1,000,000 tokens) |
| Max output | 131K |
| Input | Text · Image · Video |
| Output | Text |
| Tool calling / JSON | Yes / Yes |
| Weights & licence | Open weights · qwen3.8-max (custom) · Hugging Face |
| Parameters | 2.45T (from weight files) |
| Released | Sep 3, 2026 (OpenRouter listing date) |
| Model IDs | qwen3.8-max · qwen-3.8-max |
02Prices on 2 channels
Per 1M tokens, in the currency each source publishes (1 USD = 6.7 CNY for the blended column). Sorted: official first, then cheapest.
| Channel | Source | Input | Output | Cache hit | Blended $ | Context | Notes |
|---|---|---|---|---|---|---|---|
| Official Vendor’s China site, billed in CNY | Official 10-10 | ¥12 | ¥36 | — | 2.69 | 1M | Batch −50%Beijing region; batch -50% on supported models; explicit cache create 125% / hit 10% of input; Singapore 14.988/44.965 |
| Fireworks Inference platform (own price page) | Platform 10-10 | $2 | $6 | $0.25 | 3 | 1.05M | Batch −50%Fireworks AI own pricing page; batch -50% |
“Official” = vendor pricing page; “Platform” = the platform’s own pricing page; “OpenRouter” = price listed on OpenRouter, not checked on the platform’s site. Click a source tag to open the original page.
03Where you can use it
Derived only from the price sources above. Country restrictions are set by each vendor – check their terms before you sign up.
- Vendor’s China site, billed in CNY
Official - Inference platform (own price page)
Fireworks - Self-host the open weights
GPU / VRAM sizing and build plans→ - Free option
Qwen3.8-2.4T-A95B→
04Performance
All results →Speed and AA index from Artificial Analysis public leaderboards (not our own tests); benchmark scores from our benchmark library, best result per benchmark.
- AA indexQwen3.8 Max (0902)
- 45.4
- Output speed (median)
- 40 tok/s
- Time to first token
- 2.99s
- AA index per $ (blended)
- 16.9
| Benchmark | Best score | Source | Date |
|---|---|---|---|
| LMArena Text | 1483 | official | 2026-10-08 |
| LMArena WebDev | 1672 | official | 2026-10-08 |
| LMArena Vision | 1301 | official | 2026-10-02 |
| GPQA Diamond | 92.7% | independent | 2026-08-04 |
| AIME (OTIS Mock AIME 2024–2025) | 99.4% | independent | 2026-08-04 |
| τ²-bench | 55.2% | official | 2026-08-03 |
05Price changes we recorded
All price changes →From our daily re-checks of the sources above.
No price change recorded since we started tracking (2026-10-10).
06More from Alibaba (Qwen)
FAQ
How much does the Qwen3.8-Max API cost?
On the official pricing page: ¥12 / ¥36 per 1M input / output tokens (checked 2026-10-10).
What is the context window of Qwen3.8-Max?
1M tokens of context, up to 131K output tokens. Platforms may cap it lower – see the table.
Is Qwen3.8-Max open source?
Yes, the weights are published on Hugging Face (Qwen/Qwen3.8-2.4T-A95B). Licence: qwen3.8-max (custom).