Your sovereign, private Qwen3.8 Max
Qwen3.8-Max is the official version of the Qwen3.8 series, built on the Qwen3.8-2.4T-A95B base model with additional features including vision input, non-thinking support, a 1M default context length, and official built-in tools. A reasoning model in the ultra tier, served on ToothFairyAI's AU, EU and US endpoints — private by default, billed in US dollars per million tokens at the published rates below. No training on your data, no vendor lock-in.
How Qwen3.8 Max compares
Independent intelligence index for Qwen3.8 Max against every peer in the Ultra tier (and the tier above where needed) — same source, same tests.
Latency — end-to-end seconds for a 500-token answer
Capability indexes
Domain-weighted agentic performance for Qwen3.8 Max.
Capability index vs ultra peers
The evaluations behind the index
Independent benchmark results for Qwen3.8 Max — Elo scores as published; pass rates shown as percentages.
Per-evaluation economics
| Evaluation | Score | Time / task | Output tokens / task |
|---|---|---|---|
| AA-Briefcase | 1640.35 | 104.7 min | 334097 |
| GDPval-AA | 1667.72 | 38.3 min | 122273 |
| AutomationBench-AA | 56.2% | 6.9 min | 21920 |
| Terminal-Bench 4.0 | 38.9% | 91.7 min | 292512 |
| SciCode | 52.1% | 5.5 min | 17561 |
| Humanity's Last Exam | 43.1% | 11.6 min | 37104 |
| GDP.pdf | 22.8% | 3.5 min | 11279 |
| CritPt | 17.7% | 25.4 min | 80979 |
| AA-Omniscience | 11.98 | 0.5 min | 1462 |
| AA-LCR (long context) | 80.3% | 0.8 min | 2595 |
Run Qwen3.8 Max privately
One key, one endpoint, three jurisdictions.
- Size tierUltra
- Serving routeStandard
- Input rate$2.60/1M
- Output rate$7.80/1M
- Cached input rate$0.33/1M
- Context window1000K tokens
- Max output64K tokens
- ReasoningYes — emits thinking traces
- Vision & video inputSupports image input
- Tool callingSupported
- Data residencyAU, EU and US endpoints — processing stays inside the region you call