Alibaba · Chat
Qwen3.8 Max
Newer version: Qwen4 Max. We compared release dates and version names in our catalogue.
Trillion-scale MoE flagship for coding, long-horizon agents, and professional work, with image and video understanding across a 1M-token context.
- llminfo IQ
- 138.7
- Maker's price
- $3.00 blended / 1M
- Max context
- 1M
- Listings
- 136
Evidence and sources
International storefront (alibabacloud.com) is Singapore-incorporated; mainland operating entity 杭州阿里云智能科技有限公司 (Hangzhou).
Advantages and trade-offs
Compared with 10 chat models, picked mainly by popularity and IQ
Advantages
More provider options
Trade-offs
Costs more than most peers
Maker's price: $3.00 blended / 1M, pricier than 6 of 9 priced peers.
No open weights
The weights aren't public, so you can't run it yourself.
Input & output
as stated by the model's providersCapabilities
Supported operations and provider-stated features
Listings
| 992K | $2.00cache read $0.25 | $6.00cache write $2.50 | |
| 262K | $1.00 | $4.00 | |
| 262K | $1.15 | $3.45 | |
| 246K | $1.15 | $3.45 | |
| 1M | $1.40cache read $0.175 | $4.20 | |
| 1M | $1.56cache read $0.1946 | $4.67cache write $1.95 | |
| 1M | $1.60cache read $0.20 | $4.80 | |
| —Not published by the provider | $1.617cache read $0.2079 | $4.8626 | |
| 1M | $1.644cache read $0.206 | $4.932 | |
| 1M | $1.65cache read $0.206 | $4.951 |
Showing 10 of 125: the maker's own listing first, then the cheapest per 1M input tokens from each provider.
Also sold by 302ai, Abacus, Abliteratedllm, Above, Adzuki, Aihubmix, Aimlapi, Alibaba Cloud (China), Apertus, Cheapest Token, Chutes, Cline Pass, Cloudflare AI Gateway, Cometapi, Commandcode, Cortecs AI, Crossmodel, DeepInfra, Digitalocean, Eden AI, Empiriolabs, Eurouter, Fireworks AI, Flowdot, Gmicloud, Hugging Face, Hyper, Iteracompute, Keln, Kenari, Kilo, Llmgateway, Llmgateway Providers, Lyceum, Meganova, Merge Gateway, Modal, Nan, Novita AI, Ofox, Opencode Go, Opper, Orcarouter, Ppio, Qiniu, Requesty, Scx AI, SiliconFlow, Standardcompute, Streamlake, Teamorouter, Tensorix, Tensorx, Together AI, Tokenmix, Venice, Vercel, Wavespeed, Zenmux.
11 listings without a price are hidden.
Included in plans
Benchmarks
Named effort levels run from lowest to highest. Select a result to see its source details.
Human preference
| Benchmark | Score |
|---|---|
| Arena ELO | |
| Chinese | |
| Creative Writing | |
| English |
Show 56 more benchmarksHide 56 benchmarks
Coding
Show 2 more resultsHide 2 results
| Benchmark | Score |
|---|---|
| IOI | |
| Code Migration | |
| LiveCodeBench | |
| ProgramBench |
Show 5 more benchmarksHide 5 benchmarks
| Benchmark | Score |
|---|---|
| SkillsBench | |
| SWE-bench | |
| SWE-bench Multimodal | |
| Vibe Code Bench 1-100 | |
| Vibe Code Bench v1.1 |
Agentic
| Benchmark | Score |
|---|---|
| Tax Agent Bench | |
| Finance Agent (v2) | |
| Harvey's Legal Agent Benchmark v1 | |
| Terminal-Bench 2.1 |
Show 2 more benchmarksHide 2 benchmarks
| Benchmark | Score |
|---|---|
| Terminal-Bench 4.0 | |
| Terminal-Bench-Science |
Multilingual
Show 34 more resultsHide 34 results
Reasoning
General
| Benchmark | Score |
|---|---|
| typos | |
| AMPS_Hard | |
| connections | |
| consecutive_events |
Show 15 more benchmarksHide 15 benchmarks
| Benchmark | Score |
|---|---|
| integrals_with_game | |
| javascript | |
| logic_with_navigation | |
| olympiad | |
| paraphrase | |
| plot_unscrambling | |
| python | |
| simplify | |
| spatial | |
| story_generation | |
| summarize | |
| tablejoin | |
| tablereformat | |
| theory_of_mind | |
| typescript |
Long context
Vision
Math
Science
Instruction following
OCR
Knowledge
Factuality
Other
Show 9 more resultsHide 9 results
Show 1 more resultHide 1 result
Show 3 more resultsHide 3 results
Show 12 more resultsHide 12 results
Data credits: BenchBoard by SYuan03 (github.com/SYuan03/benchboard). MIT covers its code and data organization; no results licence is inferred. Numeric relays retain the original stating-source links. · Data by LLM Stats · Epoch AI Benchmarking Hub, CC BY 4.0 (licence) · LMArena, CC BY 4.0 (licence) · Baromètre de l’écriture IA, Johannes Eva; public numeric ranking data, © 2026 Johannes Eva, no results licence stated · BullshitBench by Peter Gostev, MIT (licence) · Keelemudelite mõõdupuu: Independent LLM leaderboard for Estonian. Eesti Keele Instituut. https://moodupuu.eki.ee, CC BY 4.0. Latest per-benchmark results selected with partial-run flags retained. (licence) · Clinical Benchmarks (clinicalbenchmarks.ai), CC BY 4.0; fractions converted to percentages; source documents retain their own terms (licence) · MineBench, MIT (licence) · Outback Yak Research, NepNewsBench v2 research draft. Public numerical aggregates from a candidate preview that the publisher says is not evidence of publication; no release licence assigned and none asserted here. · Steel.dev Agent Leaderboards, MIT (licence) · Sierra Research tau2-bench leaderboard submissions (github.com/sierra-research/tau2-bench), MIT (licence)
Around the web
- Discussion
- Intel benchmarks D270 vs. RTX PRO 6000: 1.12x throughput on Qwen3-30B-A3B (vLLM)Discussion
2 points, 2 comments
news.ycombinator.com
- Discussion
- Discussion
- Qwen3.8-Flash-Next 180B running in real time on a 16G MacBook ProDiscussion
11 points, 2 comments
news.ycombinator.com
Also known as: accounts/fireworks/models/qwen3p8-2p4t-a95b, accounts/fireworks/models/qwen3p8-max, aihubmix/qwen3.8-2.4t-a95b, aihubmix/qwen3.8-max, alibaba/qwen3.8-2.4t-a95b, alibaba/qwen3.8-max, alibaba/qwen3.8-max-0902, bailian/qwen3.8-max, bailian/qwen3.8-max-0902, cline-pass/qwen3.8-max, dashscope/qwen3.8-max, deepinfra/qwen/qwen3.8-2.4t-a95b, deepinfra/qwen/qwen3.8-max, empiriolabs/qwen3-8-max, empiriolabs/qwen3-8-max-0902, fireworks/qwen3.8-2.4t-a95b, fireworks_ai/accounts/fireworks/models/qwen3p8-max, fireworks_ai/qwen3p8-max, gmi/qwen/qwen3.8-max, gmi/qwen/qwen3.8-max-0902, modelrunner-ai/alibaba-qwen3.8-max, novita/qwen/qwen3.8-max, novita/qwen3.8-2.4t-a95b, novita/qwen3.8-max, opencode_go/qwen3.8-max, qwen 3.8 max, qwen-3-8-2-4t-a95b, qwen-3-8-max, qwen-3.8-max, qwen/qwen3.8-2.4t-a95b, qwen/qwen3.8-max, qwen/qwen3.8-max-0902, qwen/qwen3.8-max-0902:batch, qwen/qwen3.8-max:thinking, qwen3-8-2-4t-a95b, qwen3-8-max, qwen3-8-max-0902, qwen3.8-2.4t-a95b, qwen3.8-2.4t-a95b@eu, qwen3.8-max-0902, qwen3.8-max-2026-09-02, qwen3.8-max-free, qwen_ai_platform/qwen3.8-max, qwencloud/qwen3.8-max, scx-ai-gp/qwen3.8-max, scx-ai/qwen3.8-max, streamlake/qwen3.8-2.4t-a95b, together_ai/qwen/qwen3.8-2.4t-a95b
Like this one. Tap or hover any dotted value, label or icon to see where it comes from and how it is measured. open details when tapped or hovered. How this data works