TAAS Routing & Audit

How requests are mapped to the right niche and model, the empirical evidence behind those choices, and the compliance side-flow that keeps every tool routing through the model router. The audit is the product.

Snapshot 2026-07-12 05:04 UTC · auto-refreshed

98.46%
Router Compliance
10220
Niche Resolutions
1462016
Routed Calls

Empirical Model Evidence

Head-to-head results that drive niche scores — not vendor claims, actual outputs verified by TAAS.

Code-refactor routing: which model produces valid Python · coding / code_refactor · 2026-05-20

ModelResultDetail
claude-sonnet-4-6PASS2338 tokens, end_turn, valid Python
meta-llama/Llama-3.3-70B-Instruct-TurboFAILinvalid Python (unterminated string)
gpt-4oFAILinvalid Python (unterminated string)
claude-haiku-4-5FAILinvalid Python (unterminated string)

→ Re-scored the coding niche: claude-sonnet-4-6 ranked #1; the three failing models demoted. code_refactor added as a first-class scored niche.

Smart Niche Resolution

Unknown or newly-phrased niches resolved to the nearest scored niche (semantic + lexical), or flagged genuinely new and queued for benchmarking — instead of falling back to a blind default.

RequestedResolved → nicheMethodConfidenceWhen
research_synthesisresearch (research)lexical0.8172026-07-12 04:42
legal_analysisNEW — queued for scoringnew_niche2026-07-12 04:16

Model-Router Compliance Side-Flow

Every tool must route LLM/AI calls through the model router. The nightly audit detects drift, auto-fixes via the dev system, escalates what it can't fix, and quarantines as a last resort.

RunScannedCompliantViolatorsFixedEscalatedWhen
v5_audit_20260609_03020577267607119202026-06-09 03:02
v5_audit_20260608_03000876347517117002026-06-08 03:00
v5_audit_20260607_03020875867468118012026-06-07 03:02
v5_audit_20260606_03020975147396118022026-06-06 03:02
v5_audit_20260605_030247745973191402412026-06-05 03:02
v5_audit_20260601_0300066912681597202026-06-01 03:00
v5_audit_20260531_030205691167911202502026-05-31 03:02
v5_audit_20260530_030006672867217402026-05-30 03:00
v5_audit_20260529_0302126704669212932026-05-29 03:02
v5_audit_20260528_032041664666460002026-05-28 03:20
v5_audit_20260528_031508664666460002026-05-28 03:15
v5_audit_20260527_030755651665160002026-05-27 03:07
v5_audit_20260526_030714637463740002026-05-26 03:07
v5_audit_20260525_030611637463740002026-05-25 03:06

Auto-Fix Attempts

ToolOutcomeViolationsAttemptsWhen
generated.mistral_api_v1needs_escalationmodel_literal32026-07-03 03:00
case_assistant.rag_ingest_v1fixed12026-07-02 03:00
generated.mistral_api_v1needs_escalationmodel_literal32026-06-29 03:32
generated.mistral_api_v1needs_escalationmodel_literal32026-06-24 07:52
gemini.capability_verify_v1fixed12026-06-24 07:52
gemini.capability_verify_v1fixed12026-06-24 07:51
gemini.capability_verify_v1fixed12026-06-24 07:51
gemini.capability_verify_v1fixed12026-06-24 07:51
fal.threed.generate_v1fixed12026-06-23 03:01
fal.threed.cinematic_pipeline_v1fixed12026-06-21 03:00
education.vision_portal_runner_v1fixed12026-06-20 03:00
agent.auth_tiered_routerfixed12026-06-19 03:00
fal.threed.character_animation_v1fixed12026-06-16 03:01
fal.image.generate_v1fixed12026-06-15 03:00
fal.capability_verify_v1fixed12026-06-14 03:00
anthropic.catalog_ingest_v1fixed12026-06-13 03:00
account.capability_harvest_complete_v1fixed12026-06-12 03:00
anthropic.capability_verify_v1fixed12026-06-11 03:00
anthropic.capability_verify_v1needs_escalationrest_anthropic32026-06-10 03:00
trellison.agi_audit.benchmark_niche_modelsfixed22026-06-09 03:02