Qwen3.8-27B
by Alibaba · 27B parameters
Today's price · per 1M tokens
Input
$0.500
per 1M tokens
Output
$3.00
per 1M tokens
Blended
$1.13
blended $/1M — 3:1 weighted input:output
How this price is scoped: Alibaba Model Studio International list price, single 0<Token≤1M tier, read off the vendor's open-source Qwen3.8 table directly beneath the Qwen3.8-2.4T-A95B row. Alibaba marks the SKU as eligible for the context-caching discount but prints no cache rate in that table, so cached input is left unset rather than inferred — the same treatment as its Qwen3.8-2.4T-A95B sibling. Alibaba's own endpoint on OpenRouter quotes $0.425/$2.55; the number here is the maker's published rate card, not the gateway's.
Price receipt
We read Alibaba's own pricing page on Aug 31, 2026 and found qwen3.8-27b listed with a price on the same row — that read is where the number above comes from, recorded as a new model. We keep the page text we read; its content hash is 61a3d7dd95.
Overview
The 27B dense open-weights model in Alibaba's Qwen3.8 family, served by Alibaba itself on Model Studio at $0.50/$3.00 per 1M tokens — a single flat tier across the whole 1M-token context, with thinking and non-thinking modes billed alike. It undercuts the Qwen3.6-27B row on the same rate card ($0.60/$3.60) and carries roughly four times its context. Text in, text out.
Deploy this open model on rented GPUs
Open weights mean you can self-host instead of paying per-token API prices. These platforms let you serve it on demand — often cheaper at scale.
Partner links — we may earn a commission if you sign up, at no cost to you. We only list platforms we'd recommend regardless.
Capabilities
struck through = not supportedBenchmark performance
accuracy % · higher is betterWe hold no per-benchmark accuracy scores for this model yet, so it has no accuracy average. That is a gap in our coverage, not a sign the model is untested — independent evaluators often publish a composite index for a new model long before releasing its per-benchmark numbers. The percentile above is its standing across the independent composites below.
- AA Intelligence Index: 52 (Artificial Analysis composite across reasoning, knowledge and coding evals)
- AA Agentic Index: 50.9 (Tool use, planning, autonomy)
- AA-Omniscience Index: -10 (−100–100)
- GDPval-AA v2: 1543 ELO (Real-world work tasks, human baseline = 1000)
- Ranks #34 of 172 comparably-measured models by percentile score, across 4 independent measurements
- Ranks #42 of 172 comparably-tested models by normalized performance per dollar
Source: Artificial Analysis · updated Aug 31, 2026 · See full rankings →
Specifications
- Provider
- Alibaba
- Context window
- 1M tokens
- Modality
- text
- Parameters
- 27B
- Open source
- Yes — open weights available
- Released
- Aug 14, 2026
- Status
- Current
- Last updated
- Aug 28, 2026
- Tags
Availability verified: Aug 28, 2026 — listed on Alibaba's own page