# Qwen3.8 Flash Next Usage, Cost & Rank | OpenCode Data

> Qwen3.8 Flash Next costs $0.15 per 1M input tokens and $0.47 per 1M output tokens.

Open-weight experimental preview of the Qwen4 architecture: hybrid-attention MoE (125B total, 6B active) with vision encoder for coding, agent tasks, and image and video understanding

- Page: https://opencode.ai/data/alibaba/qwen3-8-flash-next
- JSON: https://opencode.ai/data/alibaba/qwen3-8-flash-next.json
- Model ID: alibaba/qwen3.8-flash-next
- Lab: Alibaba

## Model facts

| Fact | Value |
| --- | --- |
| Context window | 262K |
| Max output | 131K |
| Knowledge cutoff | - |
| Release date | 2026-08-27 |
| Input modalities | text, image, video |
| Output modalities | text |
| Reasoning | Yes |
| Tool calling | Yes |
| Open weights | Yes |
| Weights | [Hugging Face](https://huggingface.co/Qwen/Qwen3.8-Flash-Next) |

## Pricing: USD per 1M tokens

| Input | Output | Cached input | Cache write |
| --- | --- | --- | --- |
| $0.15 | $0.47 | $0.02 | - |

## Benchmarks

| Benchmark | Score | Metric | Source |
| --- | --- | --- | --- |
| DeepSWE | 58.7 | score | https://huggingface.co/Qwen/Qwen3.8-Flash-Next |
| SWE-Bench Pro | 62.5 | score | https://huggingface.co/Qwen/Qwen3.8-Flash-Next |
| SWE-Bench Multilingual | 81 | score | https://huggingface.co/Qwen/Qwen3.8-Flash-Next |
| NL2Repo | 48.1 | score | https://huggingface.co/Qwen/Qwen3.8-Flash-Next |
| CoWorkBench | 73.9 | score | https://huggingface.co/Qwen/Qwen3.8-Flash-Next |
| JobBench | 55.7 | score | https://huggingface.co/Qwen/Qwen3.8-Flash-Next |
| Agents' Last Exam | 24.3 | pass@1 | https://huggingface.co/Qwen/Qwen3.8-Flash-Next |
| Agents' Last Exam | 51.2 | score | https://huggingface.co/Qwen/Qwen3.8-Flash-Next |
| Toolathlon-Verified | 73.5 | pass@1 | https://huggingface.co/Qwen/Qwen3.8-Flash-Next |
| IFBench | 81.3 | score | https://huggingface.co/Qwen/Qwen3.8-Flash-Next |
| GPQA Diamond | 91.7 | score | https://huggingface.co/Qwen/Qwen3.8-Flash-Next |
| Humanity's Last Exam | 35.9 | score | https://huggingface.co/Qwen/Qwen3.8-Flash-Next |
| LiveCodeBench | 91.9 | score | https://huggingface.co/Qwen/Qwen3.8-Flash-Next |
| ClawEval-MM | 64.4 | pass@3 | https://huggingface.co/Qwen/Qwen3.8-Flash-Next |
| ClawEval-MM | 60.4 | average score | https://huggingface.co/Qwen/Qwen3.8-Flash-Next |
| RecreationBench | 49.9 | score | https://huggingface.co/Qwen/Qwen3.8-Flash-Next |
| AndroidWorld | 84.5 | score | https://huggingface.co/Qwen/Qwen3.8-Flash-Next |
| OSWorld | 19.4 | binary completion rate | https://huggingface.co/Qwen/Qwen3.8-Flash-Next |
| OSWorld | 52.3 | partial score | https://huggingface.co/Qwen/Qwen3.8-Flash-Next |
| Vision2Web | 64 | score | https://huggingface.co/Qwen/Qwen3.8-Flash-Next |
| ERQA | 72.3 | score | https://huggingface.co/Qwen/Qwen3.8-Flash-Next |
| LVBench | 76.6 | score | https://huggingface.co/Qwen/Qwen3.8-Flash-Next |
| RealWorldQA | 88.5 | score | https://huggingface.co/Qwen/Qwen3.8-Flash-Next |
| MathVision | 90.6 | score | https://huggingface.co/Qwen/Qwen3.8-Flash-Next |
| MathVision | 95.7 | score | https://huggingface.co/Qwen/Qwen3.8-Flash-Next |
| CharXiv Reasoning | 84.6 | score | https://huggingface.co/Qwen/Qwen3.8-Flash-Next |
| CharXiv Reasoning | 90.6 | score | https://huggingface.co/Qwen/Qwen3.8-Flash-Next |

## Methodology

- Updates: Aggregated every hour. Days and weeks use UTC.
- Tokens: Input, output, reasoning, and cached tokens for each request.
- Users and sessions: Approximate counts of distinct users and OpenCode sessions.
- Cost: Session cost is the average cost per OpenCode session. Token prices are list prices from the OpenCode model catalog.
- Retention: The share of a model's users in one week who use it again the next week.
- Citation: Cite OpenCode Data (opencode.ai/data) with the update time shown at the top of the page.