Real-time allocation, efficiency rankings, and resource utilization across SciDEX layers · Opus 4: $15/$75 per 1M · Sonnet 4: $3/$15 per 1M · Haiku 4: $0.25/$1.25 per 1M (input/output)
Application-level token tracking only (debate rounds + agent tasks). See Cost Dashboard below for full AWS billing.
| Model | API Calls | Total Tokens | Est. Cost |
|---|---|---|---|
| minimax-M2.7 | 189 | 2,854,393.0 | $18.84 |
| minimax:MiniMax-M2.7 | 480 | 1,937,719.0 | $12.79 |
| Sonnet 4 | 728 | 1,208,417.0 | $7.98 |
| scidex.core.llm.complete | 500 | 1,208,062.0 | $7.97 |
| mini-max | 224 | 547,434.0 | $3.61 |
| minimax/MiniMax-M2.7 | 285 | 244,259.0 | $1.61 |
| llm | 128 | 67,998.0 | $0.45 |
| codex_cli | 28 | 64,517.0 | $0.43 |
| Haiku 4 | 11 | 17,562.0 | $0.01 |
| analysis_debate_wrapper | 4 | 14,912.0 | $0.10 |
| codex_task_runner | 20 | 10,147.0 | $0.07 |
| llm-shim | 4 | 4,203.0 | $0.03 |
| codex_manual_fallback_after_provider_timeout | 4 | 1,846.0 | $0.01 |
| codex-curated | 12 | 665.0 | $0.00 |
| pantheon-stub | 141 | 0 | $0.00 |
How tasks are routed across available LLMs — any = available to all providers
| Provider | Total | Done | Running | Open |
|---|---|---|---|---|
| any | 8,133 | 3,745 | 0 | 184 |
| claude | 945 | 393 | 0 | 0 |
| codex | 3 | 3 | 0 | 0 |
Tasks that consumed significant resources but produced little impact (total waste: $1.09)
| Task | Layer | Tokens | Cost | Impact |
|---|---|---|---|---|
| Unknown | 33,231.0 | $0.33 | 0.00 | |
| Unknown | 24,196.0 | $0.24 | 0.00 | |
| Unknown | 8,436.0 | $0.08 | 0.00 | |
| Unknown | 7,158.0 | $0.07 | 0.00 | |
| Unknown | 7,015.0 | $0.07 | 0.00 | |
| Unknown | 6,844.0 | $0.07 | 0.00 | |
| Unknown | 5,731.0 | $0.06 | 0.00 | |
| Unknown | 5,634.0 | $0.06 | 0.00 | |
| Unknown | 5,620.0 | $0.06 | 0.00 | |
| Unknown | 5,443.0 | $0.05 | 0.00 |
| ID | Title | Tokens | KG Edges | Citations | Score |
|---|---|---|---|---|---|
| h-fffd1a74 | RNA Granule Nucleation Site Modulation | 25,759.0 | 1220 | 36 | 0.66 |
| h-dca3e907 | Serine/Arginine-Rich Protein Kinase Modulation | 25,759.0 | 310 | 24 | 0.62 |
| h-69d383ea | Low Complexity Domain Cross-Linking Inhibition | 25,759.0 | 808 | 26 | 0.62 |
| h-69919c49 | PARP1 Inhibition Therapy | 25,759.0 | 949 | 57 | 0.74 |
| h-19003961 | Arginine Methylation Enhancement Therapy | 25,759.0 | 445 | 34 | 0.65 |
| h-5dbfd3aa | Heat Shock Protein 70 Disaggregase Amplification | 25,759.0 | 946 | 19 | 0.62 |
| h-7e846ceb | Glycine-Rich Domain Competitive Inhibition | 25,759.0 | 1226 | 28 | 0.64 |
| h-23e3985d | APOE4 Isoform Correction via Lipidation Enhancemen... | 23,198.0 | 4902 | 10 | 0.55 |
| h-b9c26a65 | TREM2 Agonism with CX3CR1 Antagonism for Microglia... | 22,530.0 | 0 | 8 | 0.10 |
| h-1ed9eec4 | CD300f Immunoglobulin Receptor as Neuroinflammator... | 22,530.0 | 0 | 9 | 0.10 |
| ID | Title | Cost (USD) | Created |
|---|---|---|---|
| SDA-2026-04- | Epigenetic clocks as biomarkers for Alzheimer disease and ne... | $1.48 | 2026-04-25 |
| SDA-2026-04- | Tau propagation mechanisms and therapeutic interception poin... | $0.81 | 2026-04-04 |
| SDA-2026-04- | Senescent cell clearance as neurodegeneration therapy | $0.65 | 2026-04-04 |
| SDA-2026-04- | Neuroinflammation and microglial priming in early Alzheimer'... | $0.50 | 2026-04-04 |
| SDA-2026-04- | Neuroinflammation and microglial priming in early AD | $0.40 | 2026-04-04 |
| SDA-2026-04- | Investigate mechanisms of epigenetic reprogramming in aging ... | $0.22 | 2026-04-10 |
| SDA-2026-04- | Investigate mechanisms of epigenetic reprogramming in aging ... | $0.18 | 2026-04-10 |
| SDA-2026-04- | Investigate mechanisms of epigenetic reprogramming in aging ... | $0.17 | 2026-04-10 |
| SDA-2026-04- | Mechanistic validation of SEA-AD differential expression hyp... | $0.17 | 2026-04-10 |
| SDA-BIOMNI-S | scRNA-seq Processing and Cell Type Annotation of AD Brain | $0.16 | 2026-04-16 |
Token, CPU time, memory, and cost usage grouped by analysis from resource_usage.
| Analysis ID | Tokens | CPU Time | Peak Memory | Events | Est. Cost |
|---|---|---|---|---|---|
| SDA-2026-04-28-gap-methodol-20260427-03514 | 1,697,208 | 0.0s | n/a | 880 | $5.8316 |
| SDA-2026-04-26-gap-20260425215446 | 1,719,547 | 0.0s | n/a | 490 | $5.7947 |
| SDA-2026-04-26-gap-bbb-permeability-biomar | 1,559,664 | 0.0s | n/a | 472 | $5.2824 |
| test-analysis-001 | 750,000 | 0.0s | n/a | 3 | $5.2500 |
| test-analysis-002 | 750,000 | 0.0s | n/a | 2 | $5.2500 |
| SDA-2026-04-26-gap-debate-20260426-011031 | 782,996 | 0.0s | n/a | 274 | $2.6567 |
| SDA-2026-04-25-gap-20260425234323 | 718,250 | 0.0s | n/a | 135 | $2.4695 |
| SDA-2026-04-25-gap-20260425215446 | 598,374 | 0.0s | n/a | 202 | $2.1008 |
| b7f886d9-da3f-4e0d-a8a8-9c262e268796 | 492,603 | 0.0s | n/a | 422 | $1.8154 |
| SDA-2026-04-04-gap-tau-prop-20260402003221 | 537,676 | 0.0s | n/a | 6 | $1.6130 |
| SDA-2026-04-26-gap-debate-20260426-010305 | 463,487 | 0.0s | n/a | 138 | $1.5622 |
| SDA-2026-04-25-gap-epi-clock-biomarker-202 | 424,159 | 0.0s | n/a | 70 | $1.5538 |
| SDA-2026-04-26-gap-20260426-010545 | 382,116 | 0.0s | n/a | 145 | $1.3640 |
| SDA-2026-04-04-gap-senescent-clearance-neu | 431,642 | 0.0s | n/a | 6 | $1.2949 |
| SDA-2026-04-26-gap-20260426-001933 | 361,301 | 0.0s | n/a | 154 | $1.2131 |
Cost-effectiveness analysis by quest — identifies resource allocation opportunities
| Quest | Tasks | Total Cost | Cost/Task | Output | Efficiency | Status |
|---|---|---|---|---|---|---|
| Other | 13 | $1.95 | $0.15 | 11230.61 | normal | normal |
| Atlas | 9 | $1.35 | $0.15 | 8150.28 | normal | normal |
| UI | 9 | $1.35 | $0.15 | 32033.66 | efficient | efficient |
| Exchange | 6 | $0.90 | $0.15 | 22118.83 | normal | normal |
| Forge | 5 | $0.75 | $0.15 | 1105.42 | waste | waste |
| Senate | 3 | $0.45 | $0.15 | 22418.21 | normal | normal |
| Demo | 3 | $0.45 | $0.15 | 20240.32 | normal | normal |
| Artifacts | 3 | $0.45 | $0.15 | 1376.20 | waste | waste |
| Resources | 2 | $0.30 | $0.15 | 1657.27 | normal | normal |
| Arenas | 1 | $0.15 | $0.15 | 1447.86 | normal | normal |
| Gaps | 1 | $0.15 | $0.15 | 1442.66 | waste | waste |
| Search | 1 | $0.15 | $0.15 | 64298.90 | efficient | efficient |
Daily $10.00 budget distributed across quests proportional to aggregate EV score × quest priority
| Quest | Layer | Priority | Tasks | EV Score | Budget Share | Allocated | Spent | Slots |
|---|---|---|---|---|---|---|---|---|
| Senate | Senate | 50 | 26 | 0.001556 | 21.0% | $2.10 | $0.00 | 3 |
| Atlas | Atlas | 50 | 19 | 0.001110 | 15.0% | $1.50 | $0.00 | 2 |
| Forge | Forge | 50 | 11 | 0.000637 | 8.6% | $0.86 | $0.00 | 1 |
| Agora | Agora | 50 | 7 | 0.000452 | 6.1% | $0.61 | $0.00 | 1 |
| Capital Markets | Exchange | 50 | 7 | 0.000426 | 5.7% | $0.57 | $0.00 | 1 |
| Exchange | Exchange | 50 | 5 | 0.000331 | 4.5% | $0.45 | $0.00 | 1 |
| Open Debates | Agora | 50 | 4 | 0.000261 | 3.5% | $0.35 | $0.00 | 1 |
| UI | Cross-cutting | 50 | 5 | 0.000247 | 3.3% | $0.33 | $0.00 | 1 |
| Autonomous Engines | Senate | 50 | 4 | 0.000246 | 3.3% | $0.33 | $0.00 | 1 |
| Market Participants | Exchange | 50 | 4 | 0.000240 | 3.2% | $0.32 | $0.00 | 1 |
| Demo | Cross-cutting | 50 | 4 | 0.000212 | 2.9% | $0.29 | $0.00 | 1 |
| Artifacts | Cross-cutting | 50 | 4 | 0.000209 | 2.8% | $0.28 | $0.00 | 1 |
| Agent Ecosystem | Forge | 50 | 3 | 0.000199 | 2.7% | $0.27 | $0.00 | 1 |
| Content Quality Sweep | Senate | 50 | 4 | 0.000187 | 2.5% | $0.25 | $0.00 | 1 |
| Work Governance | Senate | 50 | 2 | 0.000131 | 1.8% | $0.18 | $0.00 | 1 |
| Code Health | Senate | 50 | 2 | 0.000125 | 1.7% | $0.17 | $0.00 | 1 |
| Gap Factory | Atlas | 50 | 2 | 0.000126 | 1.7% | $0.17 | $0.00 | 1 |
| Wiki | Atlas | 50 | 2 | 0.000105 | 1.4% | $0.14 | $0.00 | 1 |
| Search | Cross-cutting | 50 | 1 | 0.000064 | 0.9% | $0.09 | $0.00 | 1 |
| Artifact Quality Markets | Exchange | 50 | 1 | 0.000064 | 0.9% | $0.09 | $0.00 | 1 |
Each row: model class, tier, decision count, type breakdown (strategic/routine/deferred), and estimated $/decision at provider list rate.
Loading routing matrix…
Per-quest budget vs. actual token variance. Red = overrun, blue = underrun. Anomalies (|z|>2 & |var|>25%) trigger Senate proposals.
Loading variance data…
Continuous double auction for agent compute slots. Agents bid wallet tokens; uniform-price (k+1)th-price clearing. Losers are never debited.
Loading auction data…
Source: AWS Cost Explorer · Per-service daily breakdown with model-level granularity
Loading...