> ## Documentation Index > Fetch the complete documentation index at: https://docs.gdeltcloud.com/llms.txt > Use this file to discover all available pages before exploring further. # ML hardware catalog > Epoch AI ML-hardware catalog (CC-BY 4.0) — accelerators (GPU / TPU / NPU / ASIC …) with manufacturer, release date, TDP, memory, process size, and FP16/FP8 throughput. Rows sort by release date descending. Requires a plan with AI compute & models access (`can_use_epoch_ai`) — the Corporate & Supply Chain and Global Intelligence plans. Response echoes `applied_filters`. ## OpenAPI ````yaml /api-reference/openapi-v2.json get /api/v2/epoch/hardware openapi: 3.0.3 info: title: GDELT Cloud API v2 version: 2.0.0 description: >- Clean v2 REST API for generated GDELT Cloud structured Events, clustered Stories, linked Entities, summaries, admin1 discovery, significance ranking, and paginated article evidence. Event significance is a family-scoped weighted blend, renormalized so every event family spans a true 0-1: each event's raw total is divided by the maximum its own family can reach (Conflict 1.00, CAMEO+ POLITICAL 0.90, other CAMEO+ domains 0.65). All events: Goldstein severity 0.25, article evidence 0.05, confidence 0.05. Conflict only: fatalities 0.55 (log-scaled by body count) and civilian targeting 0.10. CAMEO+ only: magnitude 0.20, systemic importance 0.15, propagation potential 0.10, market sensitivity 0.10. When magnitude is unmeasured its term AND its 0.20 weight are both dropped. The four CAMEO+ metrics are rubric scores produced by published formulas - ordinal ranking signals, not measurements, probabilities, or predicted price moves. goldstein_scale is one canonical public metric, populated for Conflict Events and CAMEO+ POLITICAL Events where meaningful. Story linked_event_count is a Story-to-Event link aggregate, not a distinct Event total. servers: - url: https://gdeltcloud.com description: Production - url: http://localhost:3000 description: Local development security: - ApiKeyAuth: [] tags: - name: Events description: Coded CAMEO+ / conflict events — search, fetch, and summarize. - name: Stories description: Clustered narratives (stories) and their articles. - name: Entities description: >- People & organizations — resolve a name to an entity, then fetch its profile and tone. - name: Geography description: Admin-1 geography lookups. - name: Energy description: Global Energy Monitor assets + ownership registry. - name: Facilities description: >- Unified physical-asset directory — GEM energy assets + World Port Index ports + Epoch AI data centers on one keyed surface, resolved to spine owners. Plan-gated (Analyst plan and up). - name: Screening and Reference description: >- Restricted-party lists, screening, and ownership-chain exposure. Plan-gated (`can_use_screening` / `can_use_exposure`). - name: China description: >- China overseas development finance (AidData GCDF). Plan-gated (`can_use_china`). - name: Media Intelligence description: >- Entity tone (`can_use_tone`) and Share of Voice (`can_use_share_of_voice`) are plan-gated. Social signal remains an admin-only preview. - name: Filings description: >- SEC EDGAR filings, XBRL financials, and AI-extracted relations. Requires a plan with SEC filings access (`can_use_filings`) — the Corporate & Supply Chain plan and above. - name: Macro Finance description: >- FRED economic time series (catalog, point-in-time observations, releases). Plan-gated (`can_use_macro`). - name: Maritime description: >- AIS-derived vessel-flow signals across maritime chokepoints — chokepoint transits, last-known vessel positions, carriers, dwell, and AIS-dark gaps. Plan-gated (`can_use_maritime`). - name: Briefs description: Source-backed monitoring briefs. paths: /api/v2/epoch/hardware: get: tags: - AI Compute summary: ML hardware catalog description: >- Epoch AI ML-hardware catalog (CC-BY 4.0) — accelerators (GPU / TPU / NPU / ASIC …) with manufacturer, release date, TDP, memory, process size, and FP16/FP8 throughput. Rows sort by release date descending. Requires a plan with AI compute & models access (`can_use_epoch_ai`) — the Corporate & Supply Chain and Global Intelligence plans. Response echoes `applied_filters`. operationId: list-epoch-hardware-v2 parameters: - name: type in: query required: false schema: type: string enum: - GPU - TPU - XPU-R - NPU - GPGPU - ASIC - XPU - LPU - Hybrid CPU - DCU (GPGPU) - Other description: >- Case-insensitive exact match on the accelerator class as recorded by Epoch; the enum lists every stored type. Some tokens are composite, e.g. `DCU (GPGPU)`. An unknown value returns 400 INVALID_ENUM. - name: manufacturer in: query required: false schema: type: string description: >- Case-insensitive substring match on manufacturer (e.g. Nvidia, AMD, Google, Huawei, Cambricon). Alias `org` is accepted. - name: foundry in: query required: false schema: type: string example: SMIC description: >- Case-insensitive substring match on the fabricator. Observed values: TSMC, SMIC, Samsung, Intel (plus dual-fab `SMIC,TSMC` parts, which either `SMIC` or `TSMC` returns). Echoed in applied_filters. Not strictly validated — an unmatched value returns 0 rows rather than 400. - name: date_start in: query required: false schema: type: string format: date example: '2023-01-01' description: >- Filter to hardware with `release_date` on or after this date (YYYY-MM-DD). A malformed date is ignored rather than erroring. - name: date_end in: query required: false schema: type: string format: date example: '2024-12-31' description: >- Filter to hardware with `release_date` on or before this date (YYYY-MM-DD). Combine with `date_start` for a release-date window. - name: limit in: query required: false schema: type: integer default: 25 minimum: 1 maximum: 200 description: Rows per page (1–200). Defaults to 25. - name: cursor in: query required: false schema: type: string description: >- Canonical page token — pass the `pagination.next_cursor` from the previous response here to fetch the next page (uniform across all list endpoints). - name: offset in: query required: false schema: type: integer minimum: 0 description: >- Row offset for pagination. Prefer the canonical `cursor` token; both are supported. responses: '200': description: >- A page of ML hardware. Envelope: `{ success, data, pagination, applied_filters }`. content: application/json: schema: type: object properties: success: type: boolean data: type: array items: type: object properties: hardware_name: type: string manufacturer: type: string type: type: string release_date: type: string nullable: true release_price_usd: type: number nullable: true tensor_fp16_flops: type: number nullable: true fp8_flops: type: number nullable: true memory_bytes: type: number nullable: true tdp_w: type: number nullable: true process_size_nm: type: number nullable: true entity_id: type: string entity_page_name: type: string pagination: type: object properties: limit: type: integer offset: type: integer next_cursor: type: string nullable: true description: >- Canonical page token — pass it back as `cursor` to fetch the next page (uniform across all list endpoints). `null` when there are no more rows. next_offset: type: integer nullable: true deprecated: true description: >- Legacy — prefer `next_cursor` (still returned). Numeric row offset of the next page. applied_filters: type: object properties: manufacturer: type: string type: type: string ignored: type: object examples: example: summary: >- GET /api/v2/epoch/hardware?manufacturer=Nvidia — example response value: success: true data: - hardware_name: NVIDIA H100 SXM5 manufacturer: NVIDIA type: GPU release_date: '2022-03-22' release_price_usd: null tensor_fp16_flops: 989000000000000 fp8_flops: 1979000000000000 memory_bytes: 85899345920 tdp_w: 700 process_size_nm: 4 entity_id: e_9f8e7d6c5b4a3210 entity_page_name: Nvidia pagination: limit: 25 offset: 0 next_cursor: null next_offset: null applied_filters: manufacturer: Nvidia ignored: {} '401': description: Missing/invalid API key content: application/json: schema: $ref: '#/components/schemas/ErrorResponse' '403': description: >- Plan required (PLAN_REQUIRED) — the caller's plan does not carry the required source entitlement. content: application/json: schema: $ref: '#/components/schemas/ErrorResponse' '500': description: Server error content: application/json: schema: $ref: '#/components/schemas/ErrorResponse' security: - ApiKeyAuth: [] components: schemas: ErrorResponse: type: object properties: success: type: boolean enum: - false error: type: string code: type: string details: type: object description: >- Recoverable validation context when available, such as param, invalid_value, accepted_values, nearest_values, and a corrected example. additionalProperties: true securitySchemes: ApiKeyAuth: type: http scheme: bearer bearerFormat: gdelt_sk_... description: 'GDELT Cloud API key. Send as `Authorization: Bearer gdelt_sk_...`.' ````