DCP READ-ONLY: CONNECTED DEVELOPMENT / UI PREVIEW DCP BATCH 5/6 SOAK IN PROGRESS TELEMETRY: UNKNOWN COMMAND EXECUTION LOCKED
Model Registry
Data provenance: REAL MOCK STATIC CONFIG UNKNOWN· REAL = live read-only DCP · MOCK = demo, never production truth · STATIC CONFIG = documented defaults · UNKNOWN = not safely queryable
Runtime model assignment data is not safely queryable from read-only DCP yet. All rows below are MOCK demo placeholders and NOT current current provider assignments.
Future provider support (not configured unless verified)
OpenAI Anthropic Google Qwen local models other router providers
| Display name | Model ID | Provider | Capability | Availability | Preferred for | Used by | Fallback for | Fallback prio | Latency | Cost | Success |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Example Frontier Code | example-frontier-code | example-provider | code, reasoning, long-context | MOCK | coding, refactor, complex reasoning | developer | hermes, qa, security | 1 | 1450.0 ms | $X / $Y per Mtok (example) | 94% |
| Example Fast General | example-fast-general | example-provider | general, fast, cheap | MOCK | routing, summaries, classification | hermes | developer, devops | 2 | 420.0 ms | $X / $Y per Mtok (example) | 90% |
| Example Review Strong | example-review-strong | example-provider-2 | review, security, reasoning | MOCK | code review, security audit | qa, security | — | 2 | 1800.0 ms | $X / $Y per Mtok (example) | 92% |
| Example Ops Runner | example-ops-runner | example-provider-2 | ops, structured-output | MOCK | release prep, health checks, runbooks | devops | — | 3 | 900.0 ms | $X / $Y per Mtok (example) | 88% |
Model Status & Evidence (V1 static development registry)
Evidence is REAL_OBSERVATION only — no benchmark scores, no marketing claims. UNKNOWN stays UNKNOWN. Kimi K3 is UNAVAILABLE (HTTP 503 upstream provider failure) and is never selectable or recommended while unavailable.
| Model | Model ID | Status | Proven task classes | Unverified task classes | Known failures | Last verification | Provenance | Selectable? |
|---|---|---|---|---|---|---|---|---|
| DeepSeek V4.1 Flash | deepseek-v4.1-flash | VERIFIED | MEDIUM_FEATURE, LARGE_FEATURE, SMALL_PATCH, CODE_REVIEW, GIT_AUDIT, README_UPDATE | UI_VISION_REVIEW | — | 2026-10-05 | STATIC_CONFIG | selectable per policy |
| GLM-5.3-Flash(x) | glm-5.3-flashx | VERIFIED_FOR_MEDIUM_MULTI_FILE | GIT_AUDIT, README_UPDATE, SMALL_PATCH, MEDIUM_FEATURE | LARGE_FEATURE, SECURITY_REVIEW, UI_VISION_REVIEW | — | 2026-10-05 | STATIC_CONFIG | limited to task classes with evidence |
| Kimi K3 | kimi-k3 | UNAVAILABLE | — | README_UPDATE, GIT_AUDIT, CODE_REVIEW, SMALL_PATCH, MEDIUM_FEATURE, LARGE_FEATURE, ARCHITECTURE_REVIEW, PRD_DRAFT, PRD_CRITIC, SECURITY_REVIEW, DEVOPS_PLAN, INCIDENT_ANALYSIS, UI_VISION_REVIEW | upstream HTTP 503 on trivial prompt; provider retry failed | UNKNOWN | STATIC_CONFIG | NOT selectable — unavailable |
Try the Routing Simulator for deterministic, evidence-based recommendations and the Benchmark Scorecard for recorded task observations (read-only, static).
Model routing
Agent ≠ model. Selection is expected to become dynamic routing; no live telemetry is available yet, so fields show UNKNOWN / N/A.
| Task | Agent | Preferred | Selected | Provider | Fallback | Fallback used | Latency | Tokens | Cost | Success |
|---|---|---|---|---|---|---|---|---|---|---|
| Implement portal feature | developer | CONFIGURED LATER | UNKNOWN | UNKNOWN | CONFIGURED LATER | N/A | N/A | N/A | N/A | N/A |
| Run regression suite | qa | CONFIGURED LATER | UNKNOWN | UNKNOWN | CONFIGURED LATER | N/A | N/A | N/A | N/A | N/A |
Selection reason: “No live routing telemetry available”. Source: UNKNOWN. Model ids are not invented.
Model Routing Visibility
Visibility only. No AI router backend is implemented in this task.
- Task
- Hermes / Router
- Model selection
- Agent
- Result
Data provenance: REAL MOCK STATIC CONFIG UNKNOWN· REAL = live read-only DCP · MOCK = demo, never production truth · STATIC CONFIG = documented defaults · UNKNOWN = not safely queryable
Sample rows are MOCK.
| Task type | Complexity | Risk | Agent | Provider | Model | Selection reason | Fallback used? | Result |
|---|---|---|---|---|---|---|---|---|
| implement feature | medium | R2 | developer | example-provider | example-frontier-code | coding task, strong code reasoning preferred | no | SUCCEEDED |
| run regression | low | R1 | qa | example-provider-2 | example-review-strong | verification needs careful result checking | no | SUCCEEDED |
| release prep | medium | R2 | devops | example-provider-2 | example-ops-runner | structured runbook output preferred | yes | SUCCEEDED |
| security review | high | R2 | security | example-provider-2 | example-review-strong | adversarial/policy reasoning preferred | no | SUCCEEDED |