Your sovereign, private Qwen 3.8 2.4T A95B
Qwen3.8-2.4T-A95B is Alibaba's most capable Qwen model to date, a 2.4T-parameter sparse MoE with ~95B active parameters. It is built for autonomous, long-horizon work: multi-day coding runs, research-paper reproduction and self-improvement. 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 Qwen 3.8 2.4T A95B compares
Independent intelligence index for Qwen 3.8 2.4T A95B against every peer in the Ultra tier (and the tier above where needed) — same source, same tests.
Throughput — tokens per second, median
Latency — end-to-end seconds for a 500-token answer
Capability indexes
Domain-weighted agentic performance for Qwen 3.8 2.4T A95B.
Capability index vs ultra peers
The evaluations behind the index
Independent benchmark results for Qwen 3.8 2.4T A95B — Elo scores as published; pass rates shown as percentages.
Per-evaluation economics
| Evaluation | Score | Time / task | Output tokens / task |
|---|---|---|---|
| AA-Briefcase | 1435.33 | 54.2 min | 159322 |
| GDPval-AA | 1607.97 | 29.5 min | 86594 |
| AutomationBench-AA | 57.2% | 7.0 min | 20702 |
| Terminal-Bench 4.0 | 11.1% | 66.0 min | 193885 |
| SciCode | 54.1% | 5.3 min | 15490 |
| Humanity's Last Exam | 42.4% | 11.4 min | 33470 |
| GDP.pdf | 15.6% | 6.8 min | 19866 |
| CritPt | 20.0% | 26.8 min | 78817 |
| AA-Omniscience | 4.32 | 1.6 min | 4641 |
| AA-LCR (long context) | 80.3% | 0.5 min | 1464 |
Run Qwen 3.8 2.4T A95B privately
One key, one endpoint, three jurisdictions.
- Size tierUltra
- Serving routeStandard
- Input rate$4.20/1M
- Output rate$21.00/1M
- Cached input rate$0.53/1M
- Context window1010K tokens
- Max output120K tokens
- ReasoningYes — emits thinking traces
- Vision & video inputText only
- Tool callingSupported
- Data residencyAU, EU and US endpoints — processing stays inside the region you call