Your sovereign, private Qwen3.8 27B
Qwen3.8-27B is a distilled compact model from the Qwen3.8 family, condensing the capabilities of its larger siblings (Qwen3.8-Max and Qwen3.8-2.4T-A95B) into a fast, cost-efficient 27B package. It retains native multimodal (vision) input, function/tool calling and JSON output across a 262K-token context, making it a strong, lower-latency drop-in for agentic and browser-automation workloads. A reasoning model in the small 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 27B compares
Independent intelligence index for Qwen3.8 27B against every peer in the Small 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 27B.
Capability index vs small peers
The evaluations behind the index
Independent benchmark results for Qwen3.8 27B — Elo scores as published; pass rates shown as percentages.
Per-evaluation economics
| Evaluation | Score | Time / task | Output tokens / task |
|---|---|---|---|
| AA-Briefcase | 1403.13 | 46.9 min | 142232 |
| GDPval-AA | 1408.94 | 33.2 min | 100519 |
| AutomationBench-AA | 48.2% | 10.6 min | 32226 |
| Terminal-Bench 4.0 | 5.6% | 48.9 min | 148327 |
| SciCode | 46.6% | 6.9 min | 20839 |
| Humanity's Last Exam | 33.9% | 14.6 min | 44126 |
| GDP.pdf | 16.6% | 3.7 min | 11365 |
| CritPt | 5.4% | 35.1 min | 106387 |
| AA-Omniscience | -998.3% | 1.2 min | 3697 |
| AA-LCR (long context) | 82.0% | 0.9 min | 2809 |
Run Qwen3.8 27B privately
One key, one endpoint, three jurisdictions.
- Size tierSmall
- Serving routeStandard
- Input rate$0.17/1M
- Output rate$0.51/1M
- Cached input rate—
- Context window262K tokens
- Max output16K 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