Features Pricing About API Docs Get access →
LOCAL-FIRST LLM GATEWAY

One API key.
Every model.

Drop in Dream-Weaver and route across local GPU inference, Anthropic, OpenAI, Gemini, and a dozen others — with a single OpenAI-compatible endpoint. Your data never leaves unless you choose.

Base URL: https://api.dream-weaver.ai/v1
SCROLL
ZERO CODE CHANGES

Change one line. Get everything.

Full OpenAI API compatibility — chat completions, embeddings, streaming, function calling. Works with any SDK that speaks OpenAI.

# Before: client = OpenAI( api_key="sk-...", base_url="https://api.openai.com/v1" ) # After (route to any model, keep data local): client = OpenAI( api_key="dw-...", base_url="https://api.dream-weaver.ai/v1" ) # Same call, now routes intelligently: response = client.chat.completions.create( model="dream-weaver/auto", # or qwen3:8b, claude-sonnet-4-6, gpt-5… messages=[{"role": "user", "content": "explain transformers"}] )
qwen3:8b (local GPU) codestral:22b (local) claude-sonnet-4-6 claude-opus-5 gpt-5.4 gemini-3.1-pro deepseek-r1 dream-weaver/auto ✦
WHAT IT DOES

Built for teams who take AI seriously.

01 · Smart Routing

Every request to the right model

A mixture-of-experts classifier reads each request and routes: code to Codestral, reasoning to qwen3, long-form to Claude — automatically. Or override per-request. Cost drops, quality rises.

02 · Persistent Memory

Context that outlives the session

Semantic memory with pgvector and 1024-dim mxbai embeddings. Your codebase, your notes, your conversations — searchable across every model and every session, stored on your hardware.

03 · Privacy by default

Local first means local only — until you choose

Route sensitive workloads to your local GPU (NVIDIA or AMD). Route cost-insensitive tasks to cloud. Your data doesn't move unless the routing policy says so — you set the policy.

15+
Model backends routed
100%
OpenAI API compatible
1024
Dim semantic memory
<5ms
Routing overhead
FEATURES

Everything your AI stack needs, nothing it doesn't.

INTELLIGENT ROUTING

Mixture-of-experts dispatch, production-tested

The dream-weaver/auto model ID runs a lightweight classifier on every incoming request to determine the optimal backend: local GPU, Anthropic, OpenAI, or Ollama. Routing decisions are logged and auditable. Cost reduction with no accuracy trade-off for well-classified workloads.

$ Request: "Write a Python sort function" → classifier: code (0.91) → backend: codestral:22b (local) → latency: 8ms dispatch + inference $ Request: "Explain quantum entanglement" → classifier: reasoning (0.84) → backend: qwen3:8b (local) $ Request: "Review this 80k token contract" → classifier: long-context (0.97) → backend: claude-opus-5 (cloud)
SEMANTIC MEMORY

Persistent context across every model and session

Embeddings generated by mxbai-embed-large (1024 dims, pgvector) surface relevant memory at inference time. Every conversation, document, and code snippet stored on your hardware — not in a vendor's cloud. Searchable via the /v1/memories API or MCP tools.

# Semantic memory search results = dw.memory_search( query="k3d Ollama bridge architecture", limit=5, threshold=0.75 ) # Returns scored, ranked chunks: # [0.91] TOOLS.md — k3d Ollama bridge socat… # [0.88] memory/2026-08-09 — ollama-k3d-bridge… # [0.82] AGENTS.md — expose host GPU to k3d…
LSR — LOGICAL & SYMBOLIC REASONING

Hybrid neuro-symbolic reasoning layer

An optional reasoning pod (LSR) sits between the request and the model. For structured queries — rule-following, constraint satisfaction, fact checking — LSR augments the LLM with Prolog-style symbolic rules via inductive logic programming. Working memory persists across calls.

# Working memory — persists cross-turn POST /reasoning/working-memory { "session_id": "kris-session-042", "facts": [ "model(mi300x, amd)", "supports(mi300x, rocm6)" ] } # Query persisted logic: GET /reasoning/query?q=supports(X,rocm6) → X = mi300x ✓
PRIVACY ARCHITECTURE

Your models. Your hardware. Your rules.

Dream-Weaver runs inside your Kubernetes cluster. The routing gateway is self-hosted. Memory lives in your Postgres. Sensitive requests never leave your network unless you explicitly route them to a cloud backend.

ON-PREM INFERENCE

NVIDIA & AMD GPU support

Ollama backend supports CUDA (RTX, A100, H100, GB200) and ROCm (MI300X, MI355X). Bring your own iron, run your own models.

POLICY-GATED CLOUD

Cloud is opt-in, per-route

Define per-user, per-team, per-model-id routing policies. Compliance-sensitive traffic stays local. Everything else can use the cheapest capable model.

AUDIT TRAIL

Every request logged, nothing hidden

Full request/response audit log in Postgres. See what model answered, what it cost, how long it took. No black box vendor dashboards.

USE CASES

Who it's built for.

AI DEVELOPERS

Stop paying for tokens you could run locally

Keep your same OpenAI SDK calls. Route cheap workloads to local qwen3 or Codestral. Reserve Claude Opus for the tasks that actually need it. Watch your monthly bill drop.

ENTERPRISES & REGULATED INDUSTRIES

OpenAI compatibility, on-prem compliance

Healthcare, finance, legal — where data residency isn't optional. Deploy Dream-Weaver in your VPC. Your teams use the same developer experience, your InfoSec keeps their guarantees.

HPC & RESEARCH TEAMS

GPU clusters that do more than train

If you're running MI300X or H100 clusters, your idle inference capacity is valuable. Dream-Weaver turns your HPC investment into an internal LLM API — for your whole organization.

AI AGENT BUILDERS

Memory + reasoning for long-running agents

Persistent semantic memory, working memory via LSR, and MCP-compatible tools mean your agents remember across sessions without you building the plumbing. Focus on the agent logic.

GET STARTED

Ready to route smarter?

Dream-Weaver is currently available to early-access teams. If you're running local GPU hardware or want OpenAI-compatible access to every frontier model — get in touch.