Straits Industriesproduction
Search models, nodes, traces…⌘K
SI

Routing Policies

policy-2026-07-14

Hard placement constraints applied before any candidate is scored: regions, providers, egress trust tier, ZDR and data classification ceilings.

Route policies
5
Service profiles governed
12
ZDR-enforcing policies
3of 5
Confidential-capable
3class ≥ confidential

Route policies

A candidate endpoint must satisfy every column to survive hard filtering — rejections are recorded per candidate in the route plan.

PolicyAllowed regionsAllowed providersMin trust tierZDRData class ≤Profiles
Default enterprise routing
rp_default
ap-southeast-1us-east-1eu-west-1openaianthropicamazon-bedrockgoogle-vertexself-hosted-vllmTier Coptionalinternal7
Standard + ZDR
rp_standard_zdr
ap-southeast-1us-east-1eu-west-1openaianthropicamazon-bedrockgoogle-vertexTier Crequiredconfidential1
Confidential (SG, Tier A/B, ZDR)
rp_confidential
ap-southeast-1anthropicamazon-bedrockself-hosted-vllmTier Brequiredrestricted1
Production code paths
rp_code
ap-southeast-1us-east-1anthropicself-hosted-vllmopenaiTier Brequiredconfidential1
Media generation
rp_media
us-east-1eu-west-1openaigoogle-vertexTier Coptionalinternal2

Effective policy per profile

What a request actually runs under once its service profile resolves route, guardrail, cache and resilience policy together.

@company/general-fastlatencyv4
Placement
ap-southeast-1us-east-1eu-west-1Tier Cinternal
openaianthropicamazon-bedrockgoogle-vertexself-hosted-vllm
Guardrails
gp_baseline@3stream-inspected
5 of 5 stages enabled — Baseline safety + DLP
Cache · Resilience · Budget
cache autoproject · 300sstandard
max $0.05/req · audit metadata
Allowed models (preference order)
openai/gpt-5-minianthropic/claude-haikugoogle/gemini-3-flash
@company/general-balancedbalancedv7
Placement
ap-southeast-1us-east-1eu-west-1Tier Cinternal
openaianthropicamazon-bedrockgoogle-vertexself-hosted-vllm
Guardrails
gp_baseline@3stream-inspected
5 of 5 stages enabled — Baseline safety + DLP
Cache · Resilience · Budget
cache autoproject · 300sstandard
max $0.25/req · audit metadata
Allowed models (preference order)
anthropic/claude-sonnetopenai/gpt-5.2google/gemini-3-pro
@company/general-bestqualityv3
Placement
ap-southeast-1us-east-1eu-west-1Tier CZDRconfidential
openaianthropicamazon-bedrockgoogle-vertex
Guardrails
gp_baseline@3stream-inspected
5 of 5 stages enabled — Baseline safety + DLP
Cache · Resilience · Budget
cache autoproject · 300sstandard
max $1.50/req · audit redacted
Allowed models (preference order)
anthropic/claude-opusopenai/gpt-5.2google/gemini-3-pro
@company/code-productionqualityv9
Placement
ap-southeast-1us-east-1Tier BZDRconfidential
anthropicself-hosted-vllmopenai
Guardrails
gp_baseline@3stream-inspected
5 of 5 stages enabled — Baseline safety + DLP
Cache · Resilience · Budget
cache exact-onlyproject · 600scritical
max $0.60/req · audit redacted
Allowed models (preference order)
anthropic/claude-sonnetself-hosted/qwen3-coderopenai/gpt-5.2
@company/confidentialbalancedv6
Placement
ap-southeast-1Tier BZDRrestricted
anthropicamazon-bedrockself-hosted-vllm
Guardrails
gp_strict@5buffered-output
5 of 5 stages enabled — Strict confidential DLP
Cache · Resilience · Budget
cache offcritical
max $0.80/req · audit full
Allowed models (preference order)
anthropic/claude-sonnetanthropic/claude-haiku
@company/visionbalancedv2
Placement
ap-southeast-1us-east-1eu-west-1Tier Cinternal
openaianthropicamazon-bedrockgoogle-vertexself-hosted-vllm
Guardrails
gp_baseline@3stream-inspected
5 of 5 stages enabled — Baseline safety + DLP
Cache · Resilience · Budget
cache autoproject · 300sstandard
max $0.40/req · audit metadata
Allowed models (preference order)
google/gemini-3-proopenai/gpt-5.2anthropic/claude-sonnet
@company/embeddingscostv3
Placement
ap-southeast-1us-east-1eu-west-1Tier Cinternal
openaianthropicamazon-bedrockgoogle-vertexself-hosted-vllm
Guardrails
gp_baseline@3stream-inspected
5 of 5 stages enabled — Baseline safety + DLP
Cache · Resilience · Budget
cache autoorg · 86400sbest-effort
max $0.01/req · audit metadata
Allowed models (preference order)
openai/text-embedding-4
@company/voicelatencyv1
Placement
ap-southeast-1us-east-1eu-west-1Tier Cinternal
openaianthropicamazon-bedrockgoogle-vertexself-hosted-vllm
Guardrails
gp_baseline@3stream-inspected
5 of 5 stages enabled — Baseline safety + DLP
Cache · Resilience · Budget
cache offstandard
max $0.10/req · audit metadata
Allowed models (preference order)
openai/gpt-audio-2
@company/imagebalancedv2
Placement
us-east-1eu-west-1Tier Cinternal
openaigoogle-vertex
Guardrails
gp_baseline@3stream-inspected
5 of 5 stages enabled — Baseline safety + DLP
Cache · Resilience · Budget
cache offstandard
max $0.50/req · audit metadata
Allowed models (preference order)
openai/gpt-image-2
@company/videobalancedv1
Placement
us-east-1eu-west-1Tier Cinternal
openaigoogle-vertex
Guardrails
gp_baseline@3stream-inspected
5 of 5 stages enabled — Baseline safety + DLP
Cache · Resilience · Budget
cache offbest-effort
max $2.50/req · audit metadata
Allowed models (preference order)
openai/sora-2
@company/autobalancedv5
Placement
ap-southeast-1us-east-1eu-west-1Tier Cinternal
openaianthropicamazon-bedrockgoogle-vertexself-hosted-vllm
Guardrails
gp_baseline@3stream-inspected
5 of 5 stages enabled — Baseline safety + DLP
Cache · Resilience · Budget
cache autoproject · 300sstandard
max $0.50/req · audit metadata
Allowed models (preference order)
anthropic/claude-sonnetopenai/gpt-5.2openai/gpt-5-minianthropic/claude-haikugoogle/gemini-3-progoogle/gemini-3-flashanthropic/claude-opusself-hosted/qwen3-coder
fabric/autobalancedv5
Placement
ap-southeast-1us-east-1eu-west-1Tier Cinternal
openaianthropicamazon-bedrockgoogle-vertexself-hosted-vllm
Guardrails
gp_baseline@3stream-inspected
5 of 5 stages enabled — Baseline safety + DLP
Cache · Resilience · Budget
cache autoproject · 300sstandard
max $0.50/req · audit metadata
Allowed models (preference order)
anthropic/claude-sonnetopenai/gpt-5.2openai/gpt-5-minianthropic/claude-haikugoogle/gemini-3-progoogle/gemini-3-flashanthropic/claude-opusself-hosted/qwen3-coder

How enforcement layers compose

A request's effective constraints are the intersectionof its route policy, guardrail-issued route restrictions (e.g. a confidential-data finding forcing SG-region + Tier A/B + ZDR), the project's data-class ceiling and the profile's cost cap. Hard filters run before scoring, so a candidate that violates any layer never competes — it appears in the route plan under hardFilterRejections with the specific reason.