GPUvec
Compare GPU specs and cloud rental offers, check CUDA support, and estimate Qdrant resources.
使うべきか
品質と安全性
A
検出事項（1）
- LOWTool 'compare_accelerators' description lacks action verbcompare_accelerators 内
ツール定義とプロトコルへの準拠に関する自動分析に基づいています。
コンテキストコスト
~2,839トークン数（ツール定義）
~2.8 KB一般的なレスポンスサイズ
注意への影響は大きい（128k コンテキストの 2.22%）
これは、サーバーのツールがモデルのコンテキストに読み込まれるたびに消費されるおおよそのトークン数です。数が多いほど、ほかのタスクに使える注意が減ります。
インストール
ワンクリックインストール
これを `claude_desktop_config.json` ファイルに追加してください:
{
"mcpServers": {
"gpuvec": {
"url": "https://gpuvec.com/mcp"
}
}
}リモートエンドポイント
https://gpuvec.com/mcpstreamable-httpできること
ツール一覧
ツール（9）
🟢 読み取り専用🟡 書き込み🔴 削除⚪ 不明
🟢list_accelerators(search, manufacturer, limit, offset)
List recorded accelerator specifications, with units, field provenance and unknowns. At most 50 per page.
入力スキーマ
{
"type": "object",
"properties": {
"search": {
"default": "",
"maxLength": 120,
"pattern": "^[^\\x00-\\x1f\\x7f]*$",
"title": "Search",
"type": "string"
},
"manufacturer": {
"default": "",
"maxLength": 120,
"pattern": "^[^\\x00-\\x1f\\x7f]*$",
"title": "Manufacturer",
"type": "string"
},
"limit": {
"default": 20,
"maximum": 50,
"minimum": 1,
"title": "Limit",
"type": "integer"
},
"offset": {
"default": 0,
"maximum": 10000,
"minimum": 0,
"title": "Offset",
"type": "integer"
}
},
"title": "list_acceleratorsArguments"
}出力スキーマ
{
"type": "object",
"additionalProperties": true,
"title": "list_acceleratorsDictOutput"
}🟢get_accelerator(model)
Get one exact accelerator model or canonical slug. Unknown models are errors, not guessed specifications.
入力スキーマ
{
"type": "object",
"properties": {
"model": {
"maxLength": 120,
"minLength": 1,
"pattern": "^[^\\x00-\\x1f\\x7f]+$",
"title": "Model",
"type": "string"
}
},
"required": [
"model"
],
"title": "get_acceleratorArguments"
}出力スキーマ
{
"type": "object",
"additionalProperties": true,
"title": "get_acceleratorDictOutput"
}🟢compare_accelerators(models)
Compare 2–4 distinct exact models side by side; no invented performance score or compatibility ranking.
入力スキーマ
{
"type": "object",
"properties": {
"models": {
"items": {
"maxLength": 120,
"minLength": 1,
"pattern": "^[^\\x00-\\x1f\\x7f]+$",
"type": "string"
},
"maxItems": 4,
"minItems": 2,
"title": "Models",
"type": "array"
}
},
"required": [
"models"
],
"title": "compare_acceleratorsArguments"
}出力スキーマ
{
"type": "object",
"additionalProperties": true,
"title": "compare_acceleratorsDictOutput"
}🟢list_providers(search, limit, offset)
List GPU compute services with field-scoped official provenance, independent of priced offers and live stock.
入力スキーマ
{
"type": "object",
"properties": {
"search": {
"default": "",
"maxLength": 120,
"pattern": "^[^\\x00-\\x1f\\x7f]*$",
"title": "Search",
"type": "string"
},
"limit": {
"default": 20,
"maximum": 50,
"minimum": 1,
"title": "Limit",
"type": "integer"
},
"offset": {
"default": 0,
"maximum": 10000,
"minimum": 0,
"title": "Offset",
"type": "integer"
}
},
"title": "list_providersArguments"
}出力スキーマ
{
"type": "object",
"additionalProperties": true,
"title": "list_providersDictOutput"
}🟢find_offers(model, provider, source_backed_only, max_price_per_gpu_hour, max_instance_price_per_hour, ...)
Find recorded offers with bounded price/count/VRAM filters. Unknown units, scope or counts never pass numeric filters. Billing is a case-insensitive substring; availability is recorded, not live.
入力スキーマ
{
"type": "object",
"properties": {
"model": {
"default": "",
"maxLength": 120,
"pattern": "^[^\\x00-\\x1f\\x7f]*$",
"title": "Model",
"type": "string"
},
"provider": {
"default": "",
"maxLength": 120,
"pattern": "^[^\\x00-\\x1f\\x7f]*$",
"title": "Provider",
"type": "string"
},
"source_backed_only": {
"default": true,
"title": "Source Backed Only",
"type": "boolean"
},
"max_price_per_gpu_hour": {
"anyOf": [
{
"maximum": 1000000000,
"minimum": 0,
"type": "number"
},
{
"type": "null"
}
],
"default": null,
"title": "Max Price Per Gpu Hour"
},
"max_instance_price_per_hour": {
"anyOf": [
{
"maximum": 1000000000,
"minimum": 0,
"type": "number"
},
{
"type": "null"
}
],
"default": null,
"title": "Max Instance Price Per Hour"
},
"min_gpu_count": {
"anyOf": [
{
"maximum": 1000000,
"minimum": 1,
"type": "integer"
},
{
"type": "null"
}
],
"default": null,
"title": "Min Gpu Count"
},
"min_vram_per_gpu": {
"anyOf": [
{
"maximum": 1000000000,
"minimum": 0,
"type": "number"
},
{
"type": "null"
}
],
"default": null,
"title": "Min Vram Per Gpu"
},
"min_vram_total": {
"anyOf": [
{
"maximum": 1000000000,
"minimum": 0,
"type": "number"
},
{
"type": "null"
}
],
"default": null,
"title": "Min Vram Total"
},
"billing_model": {
"default": "",
"maxLength": 120,
"pattern": "^[^\\x00-\\x1f\\x7f]*$",
"title": "Billing Model",
"type": "string"
},
"availability": {
"default": "",
"enum": [
"",
"available",
"unavailable",
"waitlist",
"coming_soon",
"unknown"
],
"title": "Availability",
"type": "string"
},
"sort": {
"default": "price_per_gpu_hour",
"enum": [
"price_per_gpu_hour",
"-price_per_gpu_hour",
"instance_price_per_hour",
"-instance_price_per_hour"
],
"title": "Sort",
"type": "string"
},
"limit": {
"default": 20,
"maximum": 50,
"minimum": 1,
"title": "Limit",
"type": "integer"
},
"offset": {
"default": 0,
"maximum": 10000,
"minimum": 0,
"title": "Offset",
"type": "integer"
}
},
"title": "find_offersArguments"
}出力スキーマ
{
"type": "object",
"additionalProperties": true,
"title": "find_offersDictOutput"
}🟢list_price_history(model, provider, start, end, limit, ...)
Read actual approved GPU price observations, newest first. UTC dates filter capture time; distinct configurations are not one price trend.
入力スキーマ
{
"type": "object",
"properties": {
"model": {
"default": "",
"maxLength": 120,
"pattern": "^[^\\x00-\\x1f\\x7f]*$",
"title": "Model",
"type": "string"
},
"provider": {
"default": "",
"maxLength": 120,
"pattern": "^[^\\x00-\\x1f\\x7f]*$",
"title": "Provider",
"type": "string"
},
"start": {
"default": "",
"maxLength": 10,
"pattern": "^(?:\\d{4}-\\d{2}-\\d{2})?$",
"title": "Start",
"type": "string"
},
"end": {
"default": "",
"maxLength": 10,
"pattern": "^(?:\\d{4}-\\d{2}-\\d{2})?$",
"title": "End",
"type": "string"
},
"limit": {
"default": 20,
"maximum": 50,
"minimum": 1,
"title": "Limit",
"type": "integer"
},
"offset": {
"default": 0,
"maximum": 10000,
"minimum": 0,
"title": "Offset",
"type": "integer"
}
},
"title": "list_price_historyArguments"
}出力スキーマ
{
"type": "object",
"additionalProperties": true,
"title": "list_price_historyDictOutput"
}🟢compute_capability(search, status, segment, capability, limit, ...)
Retrieve source-backed NVIDIA CUDA hardware compute capability context; non-CUDA is not assigned a CC. No toolkit/framework guarantee.
入力スキーマ
{
"type": "object",
"properties": {
"search": {
"default": "",
"maxLength": 120,
"pattern": "^[^\\x00-\\x1f\\x7f]*$",
"title": "Search",
"type": "string"
},
"status": {
"default": "",
"enum": [
"",
"known",
"unknown",
"not_cuda"
],
"title": "Status",
"type": "string"
},
"segment": {
"default": "",
"enum": [
"",
"data_center",
"workstation",
"jetson",
"other"
],
"title": "Segment",
"type": "string"
},
"capability": {
"default": "",
"maxLength": 6,
"pattern": "^(?:\\d{1,2}\\.\\d{1,2})?$",
"title": "Capability",
"type": "string"
},
"limit": {
"default": 20,
"maximum": 50,
"minimum": 1,
"title": "Limit",
"type": "integer"
},
"offset": {
"default": 0,
"maximum": 10000,
"minimum": 0,
"title": "Offset",
"type": "integer"
}
},
"title": "compute_capabilityArguments"
}出力スキーマ
{
"type": "object",
"additionalProperties": true,
"title": "compute_capabilityDictOutput"
}🟢estimate_qdrant(inputs)
Calculate deterministic Qdrant capacity using the site's shared real engine. KiB inputs, byte outputs; not measured capacity or QPS sizing.
入力スキーマ
{
"type": "object",
"properties": {
"inputs": {
"$ref": "#/$defs/QdrantInputs"
}
},
"required": [
"inputs"
],
"$defs": {
"QdrantInputs": {
"additionalProperties": false,
"description": "Same field names, units and ranges as the site's shared Qdrant engine.",
"properties": {
"num_points": {
"default": 1000000,
"maximum": 1000000000000,
"minimum": 0,
"title": "Num Points",
"type": "integer"
},
"num_vectors": {
"default": 1000000,
"maximum": 1000000000000,
"minimum": 0,
"title": "Num Vectors",
"type": "integer"
},
"dimensions": {
"default": 1536,
"maximum": 65536,
"minimum": 0,
"title": "Dimensions",
"type": "integer"
},
"num_sparse": {
"default": 0,
"maximum": 1000000000000,
"minimum": 0,
"title": "Num Sparse",
"type": "integer"
},
"sparse_elements": {
"default": 0,
"maximum": 4294967295,
"minimum": 0,
"title": "Sparse Elements",
"type": "number"
},
"sparse_bytes_per_nonzero": {
"default": 8,
"maximum": 1024,
"minimum": 0.001,
"title": "Sparse Bytes Per Nonzero",
"type": "number"
},
"payload_kb": {
"default": 1,
"description": "KiB per unique point, not decimal kB",
"maximum": 1000000000,
"minimum": 0,
"title": "Payload Kb",
"type": "number"
},
"indexed_payload_kb": {
"default": 0,
"description": "KiB per unique point",
"maximum": 1000000000,
"minimum": 0,
"title": "Indexed Payload Kb",
"type": "number"
},
"datatype": {
"default": "float32",
"enum": [
"float32",
"float16",
"uint8",
"turbo16",
"turbo8",
"turbo4"
],
"title": "Datatype",
"type": "string"
},
"hnsw_m": {
"default": 16,
"maximum": 10000,
"minimum": 0,
"title": "Hnsw M",
"type": "integer"
},
"quantization": {
"default": "none",
"enum": [
"none",
"scalar",
"binary",
"binary1_5",
"binary2",
"product",
"turbo"
],
"title": "Quantization",
"type": "string"
},
"quantization_memory": {
"default": "auto",
"enum": [
"auto",
"pinned",
"cached",
"cold"
],
"title": "Quantization Memory",
"type": "string"
},
"quantization_bits": {
"default": "bits4",
"enum": [
"bits8",
"bits4",
"bits2",
"bits1_5",
"bits1"
],
"title": "Quantization Bits",
"type": "string"
},
"product_compression": {
"default": 8,
"enum": [
4,
8,
16,
32,
64
],
"title": "Product Compression",
"type": "integer"
},
"mmap": {
"default": false,
"title": "Mmap",
"type": "boolean"
},
"hnsw_on_disk": {
"default": false,
"title": "Hnsw On Disk",
"type": "boolean"
},
"payload_on_disk": {
"default": true,
"title": "Payload On Disk",
"type": "boolean"
},
"sparse_on_disk": {
"default": false,
"title": "Sparse On Disk",
"type": "boolean"
},
"cache_percent": {
"default": 0,
"maximum": 100,
"minimum": 0,
"title": "Cache Percent",
"type": "number"
},
"headroom_percent": {
"default": 20,
"maximum": 1000,
"minimum": 0,
"title": "Headroom Percent",
"type": "number"
},
"replication": {
"default": 1,
"maximum": 100000,
"minimum": 1,
"title": "Replication",
"type": "integer"
},
"shards": {
"default": 1,
"maximum": 1000000,
"minimum": 1,
"title": "Shards",
"type": "integer"
},
"nodes": {
"default": 1,
"maximum": 100000,
"minimum": 1,
"title": "Nodes",
"type": "integer"
},
"billing_hours": {
"default": 730,
"maximum": 8784,
"minimum": 0,
"title": "Billing Hours",
"type": "number"
},
"pricing_quote_id": {
"default": "custom",
"maxLength": 120,
"pattern": "^[^\\x00-\\x1f\\x7f]*$",
"title": "Pricing Quote Id",
"type": "string"
},
"hourly_per_node": {
"anyOf": [
{
"description": "User-supplied USD/node-hour; null means unknown",
"maximum": 1000000000,
"minimum": 0,
"type": "number"
},
{
"type": "null"
}
],
"default": null,
"title": "Hourly Per Node"
}
},
"title": "QdrantInputs",
"type": "object"
}
},
"title": "estimate_qdrantArguments"
}出力スキーマ
{
"type": "object",
"additionalProperties": true,
"title": "estimate_qdrantDictOutput"
}🟢list_qdrant_pricing(provider, region, limit, offset)
List dated official managed Cloud package quotes, not live prices or hardware recommendations.
入力スキーマ
{
"type": "object",
"properties": {
"provider": {
"default": "",
"maxLength": 120,
"pattern": "^[^\\x00-\\x1f\\x7f]*$",
"title": "Provider",
"type": "string"
},
"region": {
"default": "",
"maxLength": 120,
"pattern": "^[^\\x00-\\x1f\\x7f]*$",
"title": "Region",
"type": "string"
},
"limit": {
"default": 20,
"maximum": 50,
"minimum": 1,
"title": "Limit",
"type": "integer"
},
"offset": {
"default": 0,
"maximum": 10000,
"minimum": 0,
"title": "Offset",
"type": "integer"
}
},
"title": "list_qdrant_pricingArguments"
}出力スキーマ
{
"type": "object",
"additionalProperties": true,
"title": "list_qdrant_pricingDictOutput"
}コミュニティ
エビデンス
最近の観測
検証済みバージョンは記録されていませんツール 9 件