Alibaba (Qwen) · Released Aug 3, 2026Qwen3.8 Max
2.4-trillion-parameter MoE flagship for coding, professional work, multimodal understanding, and long-horizon agentic workflows.
- Capability
- 156.6ECI · #20 of 148
- Input
- $2.00per 1M tokens
- Output
- $6.00per 1M tokens
- Context
- 1M131K max output
What it costs
- Input
- $2.00per million tokens
- Output
- $6.00per million tokens
- Cached input
- $0.25per million tokens
- Cache write
- $2.50per million tokens
- Blended (3:1)
- $3.00pricier than 74% of priced models
Typical monthly bills
| Monthly usage | Estimated cost |
|---|---|
| Side project2M input + 0.5M output tokens | $7.00 |
| Team assistant25M input + 5M output tokens | $80.00 |
| Production app250M input + 50M output tokens | $800.00 |
Official Alibaba API price, as listed on models.dev.
Independent benchmarks
#20 of 148 scored models
The shaded band is Epoch AI’s confidence range (154.5–158.8); the tick marks the median scored model.
Scores from Epoch AI, run independently of Alibaba (Qwen).
The details
- Lab
- Alibaba (Qwen)
- Released
- Aug 3, 2026
- Context window
- 1,000,000 tokens
- Max output
- 131,072 tokens
- Inputs
- Text, Images, PDFs, Video
- Output
- Text
- Reasoning
- Adjustable effort low · medium · xhigh
- Tool calling
- Yes
- Structured output
- Yes
- Weights
- Proprietary
- API model ID
qwen3.8-maxon Alibaba- Availability
- 25 API providerslisted on models.dev
What Alibaba (Qwen) claims
Published by the lab at launch. Settings vary, so compare these only with care.
| Benchmark | Score | Setting | Source |
|---|---|---|---|
| Terminal-Bench v2.1 | 86.6 avg@10 | 5h timeout | Source |
| SWE-Bench Pro | 67.7 resolved | — | Source |
| DeepSWE v1.1 | 56.6 | — | Source |
| NL2Repo | 55.9 | — | Source |
| FrontierSWE | 73.5 | — | Source |
| MLS-Bench-Lite | 41 | 5h timeout | Source |
| PaperBench | 93 | Code-Dev; 3 runs; 12h timeout; Opus 4.6 judge | Source |
| AndroidBench | 75.1 avg@3 | — | Source |
| QwenSWEBench | 80.7 avg@3 | — | Source |
| QwenQoderBench | 58.4 avg@5 | — | Source |
| QwenReactBench | 1724 Elo | — | Source |
| QwenSVGBench | 1713 Elo | — | Source |
More from Alibaba (Qwen)
About Qwen3.8 Max
How much does Qwen3.8 Max cost?
Qwen3.8 Max costs $2.00 input / $6.00 output per million tokens (official Alibaba API price). Cached input is $0.25 per million tokens, and writing to the cache costs $2.50. At a 3:1 input-to-output mix that is $3.00 per million tokens, more expensive than 74% of the 360 priced models we track.
What is the context window of Qwen3.8 Max?
Qwen3.8 Max accepts up to 1,000,000 tokens per request and can write up to 131,072 tokens in one response.
How good is Qwen3.8 Max?
Epoch AI gives Qwen3.8 Max a Capabilities Index score of 156.6 (likely range 154.5–158.8), ranking it #20 of 148 models Epoch has scored. Epoch AI benchmark results: GPQA Diamond 92.7%, FrontierMath Tiers 1–3 74.7%, OTIS Mock AIME 2024–2025 99.4%, SimpleQA Verified 45.8%.
Is Qwen3.8 Max open source?
No. Qwen3.8 Max is proprietary; you use it through Alibaba’s API or partner platforms.
What inputs does Qwen3.8 Max support?
Qwen3.8 Max accepts text, images, PDFs and video and replies in text. It is a reasoning model with low, medium and xhigh effort settings, supports tool calling and can return structured JSON output.
When was Qwen3.8 Max released?
Alibaba (Qwen) released Qwen3.8 Max on Aug 3, 2026.
What are the best alternatives to Qwen3.8 Max?
The closest current models from other labs on capability, price and release date are Grok 4.6 (xAI, ECI 156.6, $2.00 / $6.00), Claude Sonnet 5 (Anthropic, ECI 156.3, $2.00 / $10.00), GLM-5.3 (Z.ai (Zhipu), ECI 155.8, $1.40 / $4.40) and Muse Spark 1.3 (Meta, ECI 156.9, $1.25 / $4.25).