Lore Hex

Model catalog

Models

Every model below is served from Lore Hex capacity under a single API contract and a single data policy. The machine-readable catalog lives at /api/v1/models.

ModelContextMax outBuilt for$ / M in$ / M out
lorehex/prometheus-2.0Prometheus 2.0 1M 32K Deep research, Long-context synthesis, Multi-step reasoning $1.20 $5.50
lorehex/liberty-2.0Liberty 2.0 256K 16K US-origin weight provenance, General reasoning, Procurement-friendly $0.60 $2.40
lorehex/openpatcher-g1OpenPatcher-G1 1M 32K Vulnerability triage, Patch generation, Whole-repo context $0.90 $3.60
lorehex/archimedes-large-1.0Archimedes-large 1.0 256K 32K Government & national security, Synthesis, Structured output $0.45 $1.40
lorehex/archimedes-small-1.0Archimedes-small 1.0 128K 16K Government & national security, High throughput, Extraction $0.03 $0.06
lorehex/glm-5.3-flashGLM 5.3 Flash 1M 128K Long-horizon agents, Coding, Cost efficiency $0.09 $0.29
lorehex/kimi-k3Kimi K3 1M 128K Agentic tool use, Long context, General reasoning $2.60 $13.00

The line

Prometheus 2.0

lorehex/prometheus-2.0

Long-horizon research and analysis model. Built for multi-step investigation over large document sets: literature sweeps, diligence packets, competitive teardowns, and synthesis across many sources in a single context.

Liberty 2.0

lorehex/liberty-2.0

General-purpose model served exclusively on US-origin open weights, on US soil. Built for buyers with procurement or regulatory constraints on model and infrastructure provenance.

OpenPatcher-G1

lorehex/openpatcher-g1

Security-focused model for vulnerability triage and remediation: reading large codebases, localizing a defect from an advisory or crash, and drafting a minimal patch with a regression test.

Archimedes-large 1.0

lorehex/archimedes-large-1.0

Large-tier model for synthesis, structured extraction, and agentic work across big document sets and repositories. Designed for government and national security applications.

Archimedes-small 1.0

lorehex/archimedes-small-1.0

Compact model for high-volume classification, extraction, and routing where cost per call and predictable throughput dominate. Designed for government and national security applications.

GLM 5.3 Flash

lorehex/glm-5.3-flash

Z.ai's GLM 5.3 Flash: a 320B mixture-of-experts model activating 18B parameters per token, with a 1M context window. Frontier-class long-horizon agent and coding work at flash-tier cost.

Kimi K3

lorehex/kimi-k3

Moonshot AI's Kimi K3, served on Lore Hex capacity under the Lore Hex data policy.

Using the API

The API is OpenAI chat-completions compatible. Point any OpenAI client at https://api.lorehex.co/v1 and pass a model id from the table above.

curl https://api.lorehex.co/v1/chat/completions \
  -H "Authorization: Bearer $LOREHEX_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"lorehex/prometheus-2.0",
       "messages":[{"role":"user","content":"Summarize this filing."}]}'

Streaming, tool calling, and JSON-mode structured output are supported on every model in the catalog.