Task Aliases

Instead of naming a provider model, you name a task. SpiderGate maps the alias to a chain of models and picks the best available one, falling back down the chain on failure.

{ "model": "spideriq/coding", "messages": [ … ] }

Send spideriq/coding and SpiderGate might serve GPT-5.5-Codex; if that's unavailable it falls to MiniMax-M2.5 — you don't change your code.

Core aliases

Live 30-day usage on the Di-Atomic workspace (a sense of what each is for in practice):

Alias

Use case

Models (in order)

30-day usage

spideriq/fast

Fastest response, simple tasks, low latency

llama-3.1-8b (Groq → NIM → Cerebras zai-glm-4.7)

5,872 req · $7.46

spideriq/extraction

Structured data / JSON output / parsing

zai-glm-4.7 → mistral-small → llama-3.1-8b → MiniMax

41,968 req · $7.83

spideriq/lead-analysis

B2B lead scoring, company vitals, CHAMP

codestral → llama-3.3-70b → gpt-oss-120b → mistral-small

40,610 req · $183.00

spideriq/creative

SEO content, multilingual copy

codestral → gpt-oss-120b → mistral-small → llama-3.3-70b

43,413 req · $165.13

spideriq/coding

Code generation, debugging, refactoring

gpt-5.5-codex → MiniMax-M2.5

30 req · $0

spideriq/research

Long-context analysis, document review

llama-3.3-70b → codestral → gpt-oss-120b → mistral-large

801 req · $0.77

spideriq/chat

General conversation, Q&A

llama-3.3-70b → mistral-small

238 req · $0.02

spideriq/translation

Localization across languages

mistral-small → llama-3.3-70b → codestral

—

spideriq/vision

Image understanding, OCR

gemini-2.5-flash → gpt-4o

—

spideriq/free

Free-tier models only, zero cost

llama-3.1-8b → mistral-small

3 req · ~$0

Alias families

  • spideriq/* — the general task pool (above).

  • agent/* — live customer-facing agents (subscription-first; e.g. agent/live-chat optimizes first-token latency).

  • opvs/* — the OPVS build factory (opvs/planner, opvs/coding, opvs/review, opvs/cheap).

Cost & PII guidance

  • Every response includes spidergate_metadata.cost_usd — the real dollar cost of that call.

  • Never send client PII on a free-tier alias (spideriq/fast, spideriq/free, spideriq/extraction) — route sensitive B2B data through spideriq/lead-analysis.

List the current aliases and their model chains at runtime → Models.