Loading the catalogue…
Open-weight Qwen MoE model with text and image input, configurable reasoning, about 6B active parameters and a 262,144-token context.
| Inference Provider | Sovereignty | Runs in | In / out per 1M | Status | Pin |
|---|---|---|---|---|---|
| L1 | €0.18 / 0.45 | Online |
31 use cases classified by the curator. The note says why.
Tool-calling, extended thinking, and 262K context directly support iterative coding agents operating over substantial repositories.
A 262K context window is well suited to grounding answers over large retrieved document sets, with extended thinking supporting cross-document reasoning.
Tool-calling and vision support fit browser interaction workflows, while extended thinking supports multi-step page-state reasoning.
Extended thinking supports tracing multi-file failure causes, while 262K context accommodates substantial logs and code context.
Vision capability supports chart interpretation, and extended thinking can connect visual trends with accompanying context.
Vision support enables reading chart content, while extended thinking can organize extracted values and relationships.
The model’s reasoning capability and large context support nuanced classification tasks, but there is no classification-specific benchmark in the profile.
The family is positioned for reasoning-heavy workloads and the model has extended thinking, but no code benchmark or coding-specific score is provided for this variant.
Tool-calling enables integrations and extended thinking supports complex support cases; the 262K context is useful for conversation and knowledge-base state.
Extended thinking and tool-calling support analytical workflows and external computation, but no data-analysis benchmark or execution environment is documented.
Vision capability supports document image interpretation, but dedicated OCR accuracy or document-parsing benchmarks are not provided.
The 262K context window supports large policy corpora, and extended thinking is useful for interpreting conditional rules; no policy-specific benchmark is provided.
Sourced numbers, not our own evaluations. Percentiles are against the whole Stav catalogue.
Computed from aa-intelligence-index, gpqa-diamond, aa-engineering-index, aa-coding-index, livecodebench, terminal-bench-v4, aa-lcr.
curl https://api.stav.ai/v1/chat/completions \
-H "Authorization: Bearer $STAV_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "qwen/qwen3-8-flash-next",
"messages": [{"role": "user", "content": "Hello!"}]
}'→ X-Stav-Request-Id · X-Stav-Model · X-Stav-Provider → X-Stav-Sovereignty-Level — the rung the request was served at