Qwen2.5 14B Instruct

LicenceFACT

apache-2.0

Fresh·Retrieved 25 Aug 2026

ParametersFACT

14.8B

Fresh·Retrieved 25 Aug 2026

Context lengthFACT

not returned by the registry

Evidence not verified

Downloads (30 days)FACT

2,840,908

Fresh·Retrieved 25 Aug 2026

LikesFACT

362

Fresh·Retrieved 25 Aug 2026

GatedFACT

no

Fresh·Retrieved 25 Aug 2026

Peak VRAM for serving this model, computed by lib/deployment/hardware.ts from weights + KV cache + 15% runtime overhead. Inputs: 14.8B parameters, 8,192 tokens of context (assumed — none fetched yet), 4 concurrent requests (assumed).

• Cloud GPU instance — 1 × NVIDIA L4 24 GB

  GPU

  NVIDIA L4 24 GB (AWS G6, Azure NVadsA10/NCads equivalents, GCP G2)

  VRAM

  24 GB

  System RAM

  64 GB

  Storage

  1000 GB

  CPU

  8–16 vCPU

  Form factor

  Cloud instance

  Indicative costUS$1 – US$2

  indicative on-demand rental, USD per hour, Aug 2026 — verify against the provider price list for your region

  The private-cloud counterpart of `onprem-small-24gb`: same model sizes, no capital outlay, and a region you choose explicitly. Running it continuously for a year usually costs more than buying the equivalent box, so it suits pilots, bursts and firms without a server room. The cloud provider becomes a data processor — a DPA and a documented region are required.

  Sizing basis

  • Weights: 14.8B parameters × 2 bytes per parameter (fp16) = 29.54 GB.
  • No model configuration was supplied, so the KV cache uses the size heuristic for this band: 0.197 MB per token, from Qwen2.5-14B-Instruct (48 × 8 × 128).
  • KV cache: 0.197 MB per token × 8,192 tokens of context × 4 concurrent requests = 6.44 GB, sized for every user holding a full context at once.
  • Overhead: 15% of weights plus KV cache for the runtime, activations and memory fragmentation = 5.4 GB.
  • GB means 10⁹ bytes, matching how GPU memory is advertised.
  • KV cache held at fp16 (2 bytes per element); quantising weights does not by itself quantise the cache.
  • Peak figure, not average: prefix caching and paged attention usually keep real usage lower, while speculative decoding, CUDA graphs and a second resident model push it higher.
  • "Concurrent" counts requests generated at the same instant, not people: 4 simultaneous requests is the worst case this estimate is built on — measure yours before buying.

• On-premise single 24 GB GPU server

  GPU

  NVIDIA RTX 4090 24 GB (or NVIDIA L4 24 GB for a rack-mounted, 72 W alternative)

  VRAM

  24 GB

  System RAM

  64 GB

  Storage

  2000 GB

  CPU

  16-core x86 server CPU (AMD EPYC 7003/9004 or Intel Xeon Scalable)

  Form factor

  Tower server

  Indicative costUS$4,000 – US$9,000

  indicative build cost for the complete machine, USD, Aug 2026 — verify with a local supplier

  The default box for a 20–60 person firm. Fits a 14B model at 4-bit with roughly 8 GB of KV cache left for concurrent chat, or an 8B model at fp16. NVMe storage sized for the model cache plus a document corpus and its embeddings. Add a UPS and an offsite backup target — this machine holds the whole knowledge base.

  Sizing basis

  • Weights: 14.8B parameters × 2 bytes per parameter (fp16) = 29.54 GB.
  • No model configuration was supplied, so the KV cache uses the size heuristic for this band: 0.197 MB per token, from Qwen2.5-14B-Instruct (48 × 8 × 128).
  • KV cache: 0.197 MB per token × 8,192 tokens of context × 4 concurrent requests = 6.44 GB, sized for every user holding a full context at once.
  • Overhead: 15% of weights plus KV cache for the runtime, activations and memory fragmentation = 5.4 GB.
  • GB means 10⁹ bytes, matching how GPU memory is advertised.
  • KV cache held at fp16 (2 bytes per element); quantising weights does not by itself quantise the cache.
  • Peak figure, not average: prefix caching and paged attention usually keep real usage lower, while speculative decoding, CUDA graphs and a second resident model push it higher.
  • "Concurrent" counts requests generated at the same instant, not people: 4 simultaneous requests is the worst case this estimate is built on — measure yours before buying.

• On-premise single 48 GB GPU server

  GPU

  NVIDIA L40S 48 GB (rack) or NVIDIA RTX 6000 Ada 48 GB (office workstation)

  VRAM

  48 GB

  System RAM

  128 GB

  Storage

  4000 GB

  CPU

  24–32-core x86 server CPU (AMD EPYC or Intel Xeon Scalable)

  Form factor

  Rack server

  Indicative costUS$12,000 – US$22,000

  indicative build cost for the complete machine, USD, Aug 2026 — verify with a local supplier

  The step up when answer quality matters more than price: a 32B model at 4-bit with generous KV cache, or a 14B model at fp16 with a long context. ECC memory and a standard 2U chassis mean it belongs in a rack or comms room rather than under a desk.

  Sizing basis

  • Weights: 14.8B parameters × 2 bytes per parameter (fp16) = 29.54 GB.
  • No model configuration was supplied, so the KV cache uses the size heuristic for this band: 0.197 MB per token, from Qwen2.5-14B-Instruct (48 × 8 × 128).
  • KV cache: 0.197 MB per token × 8,192 tokens of context × 4 concurrent requests = 6.44 GB, sized for every user holding a full context at once.
  • Overhead: 15% of weights plus KV cache for the runtime, activations and memory fragmentation = 5.4 GB.
  • GB means 10⁹ bytes, matching how GPU memory is advertised.
  • KV cache held at fp16 (2 bytes per element); quantising weights does not by itself quantise the cache.
  • Peak figure, not average: prefix caching and paged attention usually keep real usage lower, while speculative decoding, CUDA graphs and a second resident model push it higher.
  • "Concurrent" counts requests generated at the same instant, not people: 4 simultaneous requests is the worst case this estimate is built on — measure yours before buying.

• Apple Silicon workstation (unified memory)

  GPU

  Apple M4 Max (up to 128 GB unified) or M3 Ultra (up to 256 GB unified)

  VRAM

  96 GB

  System RAM

  128 GB

  Storage

  2000 GB

  CPU

  Apple M4 Max / M3 Ultra

  Form factor

  Workstation

  Indicative costUS$4,500 – US$12,000

  indicative build cost for the complete machine, USD, Aug 2026 — verify with a local supplier

  Quiet, standard-socket power draw, and a large memory pool — about 75% of installed unified memory is addressable by the GPU, which is where the 96 GB figure comes from on a 128 GB M4 Max. Memory bandwidth, not capacity, sets the ceiling: excellent for a pilot, a partner workstation or a single heavy user, weak for 40 people at once. Runs Ollama, llama.cpp and MLX; does not run CUDA builds of vLLM.

  Sizing basis

  • Weights: 14.8B parameters × 2 bytes per parameter (fp16) = 29.54 GB.
  • No model configuration was supplied, so the KV cache uses the size heuristic for this band: 0.197 MB per token, from Qwen2.5-14B-Instruct (48 × 8 × 128).
  • KV cache: 0.197 MB per token × 8,192 tokens of context × 4 concurrent requests = 6.44 GB, sized for every user holding a full context at once.
  • Overhead: 15% of weights plus KV cache for the runtime, activations and memory fragmentation = 5.4 GB.
  • GB means 10⁹ bytes, matching how GPU memory is advertised.
  • KV cache held at fp16 (2 bytes per element); quantising weights does not by itself quantise the cache.
  • Peak figure, not average: prefix caching and paged attention usually keep real usage lower, while speculative decoding, CUDA graphs and a second resident model push it higher.
  • "Concurrent" counts requests generated at the same instant, not people: 4 simultaneous requests is the worst case this estimate is built on — measure yours before buying.

No hosted provider has been fetched for this model. OpenRouter’s catalogue is read by pnpm sync; until it runs, this is empty rather than assumed.

• Self hostedVoice agent (telephony + STT + LLM + TTS)LiveKit Agents or Pipecat as the real-time orchestrator, a SIP trunk for telephony, faster-whisper for speech to text, a local model for the conversation, and a self-hosted TTS voice. Every hop counts against a latency budget of roughly a second.
• Self hostedDocument classification and OCR pipelineDocling (or Unstructured) converts and OCRs, embeddings plus rules classify, a local model reads only the documents that need reading, and Paperless-ngx gives the business a place to search and file. Starts CPU-only; a GPU is added when the volume justifies it.
• Self hostedPrivate company knowledge base (self-hosted RAG)Open WebUI as the employee interface, vLLM serving a Qwen2.5-14B-Instruct model on a single 24 GB GPU, PostgreSQL with pgvector for chats and embeddings, and OIDC single sign-on — all in Docker on one server in your office or colocation rack. Ollama replaces vLLM for teams under about 20 users; a 48 GB GPU lets you run a 32B model for better answers.
• Self hostedMeeting and call transcription to CRM notesfaster-whisper or WhisperX transcribes and separates speakers on your own GPU, a local model writes the summary and the action items against a fixed schema, and n8n files the result against the right CRM record with a person approving before it is saved.
• Self hostedLocal LLM inference server (Ollama / vLLM)vLLM for throughput or Ollama for simplicity, one GPU, and LiteLLM in front as the gateway that issues per-team keys and records spend. The endpoint other recipes point at.
• Self hostedPrivate ChatGPT on your own serverOllama serving an open-weight model on one GPU, Open WebUI in front of it for accounts, chats and admin controls, both in Docker on a single machine. Deliberately the smallest thing that works: no vector database, no connectors, no cluster.

1. 01–07
   Tier 4·official repository / model cardHugging Face Hub

   Hugging Face model card data — Qwen/Qwen2.5-14B-Instruct

   https://huggingface.co/api/models/Qwen/Qwen2.5-14B-Instruct

   Fresh·Retrieved 25 Aug 2026sha256:f047cd241b53

   7 records01 likes ·02 license ·03 library name ·04 lastModified ·05 gated ·06 total ·07 downloads