Qwen3.8 2.4T A95B (NVFP4)
by Alibaba (Qwen)Chat
Open-weight sparse MoE (2.4T total, 95B active), the open-weight twin of Qwen3.8 Max for coding, research, complex reasoning, and agentic workflows
- Context
- 1M tokens
- Max output
- 131,072 tokens
- Parameters
- 2.45T total, 95B active
- Released
- 2026-08-12
- Weights
- Open
- Licence
- qwen3.8-max
alibaba/qwen3-8-2-4t-a95b · also seen as Inferact/Qwen3.8-2.4T-A95B-NVFP4, accounts/fireworks/models/qwen3p8-2p4t-a95b, Qwen/Qwen3.8-2.4T-A95B, qwen-3-8-2-4t-a95b, +4 more
Specifications
Full facts as recorded, incl. what is unknown.
- Released
- 2026-08-12
- Weights
- Opennot gated
- License
- qwen3.8-max
- Input
- Text
- Output
- Text
- Context window
- 1,000,000 max across endpoints
- Max output
- 131,072
- Providers
- 13 + 12 gateways, not counted as providers
- Epoch Capabilities Index
- not evaluated
- Direct price range
- $1.95–$3.44 in · $5.70–$8.60 out per 1M tokens · 12 of 13 providers have a comparable price
More details
- Last updated
- 2026-08-26
- Knowledge cutoff
- Weights link
- Hugging Face · Qwen/Qwen3.8-2.4T-A95B
- Parameters
- 2.45T total, 95B active
- Developer country
- China
Capabilities
Across all endpoints: supported if any endpoint supports it.
- Tool calling
- Parallel toolsunknown
- Structured output
- Reasoning
- Image input
- PDF input
- Audio input
- Audio outputunknown
- Video input
- Prompt caching
- Web search
- Computer useunknown
- System messagesunknown
- Assistant prefillunknown
- Temperature
- Streamingunknown
- Attachments
Endpoints
One row per provider and variant; gateways are listed separately.
Loading endpoints…
Price history
Atom feedNo change since 19 Oct 2026. The cheapest direct price has been $1.95 in / $5.70 out per 1M tokens at IteraCompute on every data refresh since then.
Tracked from 19 Oct 2026. Headline-eligible direct prices only, so a price the sanity check flags never draws a spike. Series as JSON
Cost calculator
Loading endpoints…
Similarly priced models
Blended rate per 1M tokens, each at its own cheapest provider (10M input : 2M output mix).
| Model | Blended/1M | In/1M | Out/1M | Context | vs Qwen3.8 2.4T A95B (NVFP4) |
|---|---|---|---|---|---|
| Qwen3.8 Max 0902Alibaba (Qwen) | $2.67 | $2.00 | $6.00 | 1M | +0% |
| Mistral LargeMistral AI | $2.67 | $2.00 | $6.00 | 131K | +0% |
| Mistral Large 2.1Mistral AI | $2.67 | $2.00 | $6.00 | 131K | +0% |
| Mixtral 8x22B InstructMistral AI | $2.67 | $2.00 | $6.00 | 65.5K | +0% |
| Pixtral LargeMistral AI | $2.67 | $2.00 | $6.00 | 128K | +0% |
| Fugu MaxSakana AI | $2.67 | $2.00 | $6.00 | 1M | +0% |
| Grok LatestxAI | $2.67 | $2.00 | $6.00 | 500K | +0% |
| Grok 4.20 Beta Non-Reasoning (0309)xAI | $2.67 | $2.00 | $6.00 | 1M | +0% |
More from Alibaba (Qwen)
Newest Alibaba (Qwen) models in the catalogue. All Alibaba (Qwen) models
| Model | Released | Context | Listed price |
|---|---|---|---|
| Qwen Image 2.1 Pro | 2026-10-16 | 65.5K | unpriced |
| Qwen 3.8 27B Fimi | 2026-10-05 | 262K | $0.10 via NanoGPT |
| Qwen 3.8 27B Kiwen | 2026-10-05 | 262K | $0.10 via NanoGPT |
| Qwen 3.8 27B Swift | 2026-10-05 | 262K | $0.10 via NanoGPT |
| Qwen 3.8 27B Kimiko | 2026-10-05 | 262K | $0.10 via NanoGPT |
| Qwen 3.8 27B Synthia | 2026-10-05 | 262K | $0.10 via NanoGPT |
| Qwen 3.8 27B Antislop | 2026-10-05 | 262K | $0.10 via NanoGPT |
| Qwen 3.8 27B DarkIdol | 2026-10-05 | 262K | $0.10 via NanoGPT |