Add the open-weight vendor hosted API reference (DeepSeek, Qwen, Kimi, GLM)

CLAUDE.md

@@ -62,6 +62,8 @@ cloud-finops-skills/
 │       ├── finops-ai-value-management.md   <- AI Investment Council, stage gates
 │       ├── finops-genai-capacity.md        <- Provisioned vs shared capacity
 │       ├── finops-ai-self-hosted-vs-managed.md  <- Self-host vs managed inference
+│       ├── finops-open-weight-vendors.md    <- Open-weight vendor hosted APIs
+│       │                                     (DeepSeek, Qwen, Kimi, GLM)
 │       ├── finops-ai-dev-tools.md          <- Cursor / Claude Code / Copilot / Windsurf / Codex
 │       ├── finops-anthropic.md             <- Anthropic billing
 │       ├── finops-aws.md                   <- AWS FinOps core (billing data, rightsizing,
@@ -100,7 +102,7 @@ cloud-finops-skills/
 │                             faceted retrieval over references + playbooks;
 │                             stdio + streamable-HTTP transports; MCP Apps
 │                             resources: playbook-viewer, playbook-explorer,
-│                             reference-browser - shared JS in ui/_*.js is
+│                             reference-browser - shared JS/CSS in ui/_* is
 │                             inlined at import by server._load_ui)
 ├── scripts/               <- `fcp-coverage.sh` and `playbook-coverage.sh`
 │                             (parse frontmatter, emit the two coverage
@@ -116,7 +118,8 @@ cloud-finops-skills/
 ├── fcp-coverage.md        <- Generated FCP coverage matrix (22 caps; CI-gated)
 ├── playbook-coverage.md   <- Generated waste-playbook matrix (category x
 │                             scope, gaps listed; CI-gated)
-├── .gitattributes         <- Force LF on *.sh for Windows checkouts
+├── .gitattributes         <- Force LF on *.sh and fcp-coverage.md (Windows
+│                             checkouts with core.autocrlf=true)
 └── pipeline/              <- Content update pipeline (gitignored, private)
     ├── run_scan.py        <- Fortnightly scan entry point
     ├── run_apply.py.FROZEN <- Review and apply entry point, frozen since the
@@ -415,9 +418,10 @@ connector on claude.ai): tools executed, the host reserved the widget
 iframe, the frame stayed blank - the predicted failure. `ui.domain` was
 therefore applied on 2026-08-20: the three widget resources declare
 `meta={"ui": {"domain": ...}}` where the value is DERIVED in server.py
-from `CANONICAL_CONNECTOR_URL` (sha256 of the URL exactly as users enter
-it, no trailing slash, first 32 hex chars + `.claudemcpcontent.com`) and
-pinned by tests. When touching this, never hash an internal path instead
+from `CANONICAL_CONNECTOR_URL` (sha256 of the ROOT connector URL exactly as
+users enter it, **trailing slash included** - `CANONICAL_CONNECTOR_ORIGIN + "/"`
+in server.py - first 32 hex chars + `.claudemcpcontent.com`) and pinned by
+tests. When touching this, never hash an internal path instead
 of the public URL - that wrong-input variant is exactly what broke
 ai-pricing-hub-mcp. Correction to an earlier note in this entry:
 ai-pricing-hub was NOT fixed on 2026-08-19 - no such fix exists in that
@@ -462,8 +466,13 @@ in a file.
 and the shape of a break-even calculation are durable, and they are what a practitioner
 actually reasons with - so they stay. Absolute prices route to a live source (OptimToken),
 and where a worked example genuinely needs a number, it carries an inline as-of date.
-After the purge, `grep -E '\$[0-9.]+\s*(/|per )\s*(1M|MTok)'` over `references/` returns
-nothing.
+Immediately after the purge, `grep -E '\$[0-9.]+\s*(/|per )\s*(1M|MTok)'` over
+`references/` returned nothing. It is no longer empty and is not meant to be: the rule
+permits a dated, explicitly-illustrative figure where a worked example needs one, and
+`finops-open-weight-vendors.md` (August 2026) carries several, because the argument it
+makes - that an open-weight flagship can price at or above a Western mid-tier model - is
+not expressible as a ratio. Treat the grep as a review prompt, not a pass/fail gate:
+every hit it returns must carry an as-of date and be marked illustrative.
 
 Two things this bought beyond correctness: the answer now carries its own date and source,
 which is what makes it usable in a client deliverable; and the pipeline's rotating
@@ -533,7 +542,7 @@ discount mechanics.
 
 ## How to add a new reference file
 
-Follow these five steps whenever you add a new domain:
+Follow these six steps whenever you add a new domain:
 
 1. **Create the reference file** in `skills/cloud-finops/references/`
    - Name it `finops-{domain}.md` (or `{category}-{domain}.md` for non-FinOps topics like `greenops-cloud-carbon.md`)
@@ -556,7 +565,19 @@ Follow these five steps whenever you add a new domain:
    - Add the file to the "Directory structure" listing
    - Add usage examples if applicable
 
-5. **Do NOT bump versions in the content PR (release-train rule, 2026-08)**
+5. **Register the file in the three CI-gated places.** Skipping any of these
+   fails the `CI` workflow, so they are not optional tidying:
+   - `llms.txt` References section (gated by `scripts/check-llms-txt.sh`)
+   - The CLAUDE.md "Repository structure" tree above (gated by
+     `scripts/check-docs-drift.sh`)
+   - `install.sh` per-tool routing: the ChatGPT inline routing table and the
+     Gemini grouped-knowledge `cat_required` list
+
+   Then regenerate the FCP coverage matrix and its heat map, which change
+   whenever a new file's frontmatter lands: `./scripts/fcp-coverage.sh` and
+   `python scripts/render-fcp-heatmap.py` (both `--check` gated in CI).
+
+6. **Do NOT bump versions in the content PR (release-train rule, 2026-08)**
    - Content PRs never touch `.claude-plugin/plugin.json`,
      `.claude-plugin/marketplace.json` versions, or `mcp_server/pyproject.toml`.
      Every `plugin.json` bump that reaches main triggers one tag + GitHub
@@ -765,7 +786,7 @@ Two standing rules that follow from the map:
       phrasing. `llms.txt` is the one list to keep current.)
 - [ ] **Content PR: no version bump.** `.claude-plugin/plugin.json`,
       `.claude-plugin/marketplace.json` `metadata.version`, and
-      `mcp_server/pyproject.toml` are untouched (release-train rule - see step 5
+      `mcp_server/pyproject.toml` are untouched (release-train rule - see step 6
       of "How to add a new reference file"). A `plugin.json` bump reaching main
       publishes to PyPI, so bumps live only in dedicated release PRs.
 - [ ] **Release PR only: bump all four versions together** - `plugin.json`

README.md

@@ -147,6 +147,9 @@ The skill provides accurate, framework-aligned guidance across the following dom
   spillover mechanics, throughput units, cross-provider comparison
 - **Self-hosted vs managed AI inference** - decision framework for self-host vs managed LLM,
   hidden cost surface, ML-Ops maturity rubric, hybrid routing patterns (LiteLLM, Portkey)
+- **Open-weight vendor hosted APIs** - DeepSeek, Qwen, Kimi and GLM sold direct: the three
+  buying channels for one checkpoint, time-of-day pricing, cache and batch mechanics,
+  licensing as a cost input, data residency as the channel selector
 - **Anthropic billing** - Claude model pricing, Fast mode, long-context cliffs,
   prompt caching, Batch API, governance controls
 - **AWS Bedrock** - model pricing, provisioned throughput, batch inference, cost allocation
@@ -334,6 +337,7 @@ cloud-finops-skills/
     │   ├── finops-ai-value-management.md       ← AI investment governance
     │   ├── finops-genai-capacity.md            ← GenAI capacity models (cross-provider)
     │   ├── finops-ai-self-hosted-vs-managed.md ← Self-hosted vs managed AI inference decision
+    │   ├── finops-open-weight-vendors.md    ← Open-weight vendor hosted APIs (DeepSeek, Qwen, Kimi, GLM)
     │   ├── finops-anthropic.md                 ← Anthropic billing + governance
     │   ├── finops-aws.md                       ← AWS FinOps core
     │   ├── finops-aws-commitments.md           ← AWS SPs / RIs / Spot, liquidity, EDP

fcp-coverage.md

@@ -30,9 +30,9 @@ Marker legend:
 
 ## Optimize Usage & Cost
 
-- [x] **Architecting & Workload Placement** - finops-agentic.md; finops-ai-self-hosted-vs-managed.md; finops-onboarding-workloads.md _(secondary: finops-aws-commitments.md; finops-aws-patterns.md; finops-azure-commitments.md; finops-azure-patterns.md; finops-genai-capacity.md; finops-kubernetes.md; finops-waste-detection-playbooks.md; greenops-cloud-carbon.md)_
-- [x] **Usage Optimization** - finops-ai-dev-tools.md; finops-anthropic.md; finops-aws-patterns.md; finops-azure-patterns.md; finops-bedrock.md; finops-databricks.md; finops-snowflake.md; finops-waste-detection-playbooks.md _(secondary: finops-agentic.md; finops-ai-self-hosted-vs-managed.md; finops-aws.md; finops-azure-openai.md; finops-azure.md; finops-fabric.md; finops-for-ai.md; finops-gcp.md; finops-genai-capacity.md; finops-kubernetes.md; finops-oci.md; finops-vertexai.md; greenops-cloud-carbon.md)_
-- [x] **Rate Optimization** - finops-aws-commitments.md; finops-aws.md; finops-azure-commitments.md; finops-azure-openai.md; finops-azure.md; finops-fabric.md; finops-gcp.md; finops-genai-capacity.md; finops-vertexai.md _(secondary: finops-ai-self-hosted-vs-managed.md; finops-bedrock.md; finops-databricks.md; finops-snowflake.md)_
+- [x] **Architecting & Workload Placement** - finops-agentic.md; finops-ai-self-hosted-vs-managed.md; finops-onboarding-workloads.md _(secondary: finops-aws-commitments.md; finops-aws-patterns.md; finops-azure-commitments.md; finops-azure-patterns.md; finops-genai-capacity.md; finops-kubernetes.md; finops-open-weight-vendors.md; finops-waste-detection-playbooks.md; greenops-cloud-carbon.md)_
+- [x] **Usage Optimization** - finops-ai-dev-tools.md; finops-anthropic.md; finops-aws-patterns.md; finops-azure-patterns.md; finops-bedrock.md; finops-databricks.md; finops-snowflake.md; finops-waste-detection-playbooks.md _(secondary: finops-agentic.md; finops-ai-self-hosted-vs-managed.md; finops-aws.md; finops-azure-openai.md; finops-azure.md; finops-fabric.md; finops-for-ai.md; finops-gcp.md; finops-genai-capacity.md; finops-kubernetes.md; finops-oci.md; finops-open-weight-vendors.md; finops-vertexai.md; greenops-cloud-carbon.md)_
+- [x] **Rate Optimization** - finops-aws-commitments.md; finops-aws.md; finops-azure-commitments.md; finops-azure-openai.md; finops-azure.md; finops-fabric.md; finops-gcp.md; finops-genai-capacity.md; finops-open-weight-vendors.md; finops-vertexai.md _(secondary: finops-ai-self-hosted-vs-managed.md; finops-bedrock.md; finops-databricks.md; finops-snowflake.md)_
 - [x] **Licensing & SaaS** - finops-itam.md; finops-sam.md _(secondary: finops-ai-dev-tools.md)_
 - [x] **Sustainability** - greenops-cloud-carbon.md
 

install.sh

@@ -627,6 +627,7 @@ Use these knowledge files for the following query types:
 | AI investment governance, Investment Council, stage gates, AI business case, value quantification methods, ROI sensitivity | finops-ai-value-management.md |
 | GenAI capacity planning, provisioned vs shared, spillover | finops-genai-capacity.md |
 | Self-hosted vs managed AI inference, build-vs-buy LLM, vLLM, GPU rental, hidden cost surface | finops-ai-self-hosted-vs-managed.md |
+| Open-weight vendor hosted APIs (DeepSeek, Qwen, Kimi, GLM), time-based pricing, open-weight licensing | finops-open-weight-vendors.md |
 | AI coding tools (Cursor, Copilot, Claude Code, Codex, Windsurf, Gemini Code Assist) | finops-ai-dev-tools.md |
 | Databricks (system.billing.usage, DBCU, allocation, Photon) | finops-databricks.md |
 | Microsoft Fabric (F-SKUs, CU smoothing, pause/resume, governance trap) | finops-fabric.md |
@@ -750,6 +751,7 @@ build_gemini_grouped_knowledge() {
     "$refs/finops-for-ai.md" "$refs/finops-anthropic.md" "$refs/finops-ai-dev-tools.md" \
     "$refs/finops-genai-capacity.md" "$refs/finops-ai-value-management.md" \
     "$refs/finops-ai-self-hosted-vs-managed.md" \
+    "$refs/finops-open-weight-vendors.md" \
     "$refs/finops-agentic.md"
   cat_required "$outdir/data-platforms.md" \
     "$refs/finops-databricks.md" "$refs/finops-fabric.md" "$refs/finops-snowflake.md"

llms.txt

@@ -29,6 +29,7 @@ Hub at https://optimtoken.optimnow.io.
 - [finops-ai-value-management.md](https://raw.githubusercontent.com/OptimNow/cloud-finops-skills/main/skills/cloud-finops/references/finops-ai-value-management.md): AI investment governance, stage gates
 - [finops-genai-capacity.md](https://raw.githubusercontent.com/OptimNow/cloud-finops-skills/main/skills/cloud-finops/references/finops-genai-capacity.md): Provisioned vs shared capacity, cross-provider
 - [finops-ai-self-hosted-vs-managed.md](https://raw.githubusercontent.com/OptimNow/cloud-finops-skills/main/skills/cloud-finops/references/finops-ai-self-hosted-vs-managed.md): Self-hosted vs managed AI inference decisioning
+- [finops-open-weight-vendors.md](https://raw.githubusercontent.com/OptimNow/cloud-finops-skills/main/skills/cloud-finops/references/finops-open-weight-vendors.md): Open-weight vendor hosted APIs (DeepSeek, Qwen, Kimi, GLM) - channels, rate mechanics, licensing
 - [finops-ai-dev-tools.md](https://raw.githubusercontent.com/OptimNow/cloud-finops-skills/main/skills/cloud-finops/references/finops-ai-dev-tools.md): AI coding tools (Cursor, Claude Code, Copilot, Codex, Windsurf)
 - [finops-anthropic.md](https://raw.githubusercontent.com/OptimNow/cloud-finops-skills/main/skills/cloud-finops/references/finops-anthropic.md): Anthropic/Claude billing and governance
 - [finops-aws.md](https://raw.githubusercontent.com/OptimNow/cloud-finops-skills/main/skills/cloud-finops/references/finops-aws.md): AWS FinOps (EC2, CUR, Savings Plans, cost allocation)

skills/cloud-finops/POWER.md

@@ -35,6 +35,13 @@ keywords:
   - ptu
   - vertex ai
   - anthropic billing
+  - deepseek
+  - qwen
+  - kimi
+  - glm
+  - open-weight pricing
+  - chinese model apis
+  - time-based pricing
   - claude pricing
   - genai capacity
   - provisioned throughput
@@ -263,6 +270,7 @@ methodology questions - not for routine billing-mechanics queries.
 | AI investment governance, AI Investment Council, stage gates, incremental funding, AI value management, AI practice operations, AI business case, quantifying AI business value, cost displacement vs revenue uplift vs retention vs premium monetisation, realisation rate, ROI sensitivity analysis, payback, break-even volume | `references/finops-ai-value-management.md` |
 | GenAI capacity planning, provisioned vs shared capacity, traffic shape, spillover, throughput units | `references/finops-genai-capacity.md` |
 | Self-hosted vs managed AI inference, build vs buy LLM, vLLM, SGLang, llama.cpp, GPU rental, RunPod, CoreWeave, Lambda, hidden cost surface, ML-Ops maturity rubric, hybrid routing (LiteLLM, Portkey) | `references/finops-ai-self-hosted-vs-managed.md` |
+| Open-weight model vendors on their own hosted APIs, DeepSeek, Qwen, Kimi, Moonshot, GLM, Z.ai, Chinese model APIs, open-weight pricing, time-based pricing, peak and off-peak token rates, vendor API vs third-party host channel choice, open-weight model licensing, GLM Coding Plan | `references/finops-open-weight-vendors.md` |
 | AWS billing data, CUR, Data Exports for FOCUS 1.2, Cost Explorer, EC2/compute rightsizing, SageMaker operational FinOps, cost allocation, governance, CloudFront flat-rate plans, S3 Files, multi-org billing | `references/finops-aws.md` |
 | AWS Savings Plans, Reserved Instances, Spot, commitment decision tree, commitment portfolio liquidity, phased purchasing, EDP negotiation, Convertible RI exchange | `references/finops-aws-commitments.md` |
 | AWS per-service inefficiency catalogue, enumerated AWS optimisation patterns | `references/finops-aws-patterns.md` |

skills/cloud-finops/SKILL.md

@@ -38,6 +38,7 @@ methodology questions - not for routine billing-mechanics queries.
 | AI investment governance, AI Investment Council, stage gates, incremental funding, AI value management, AI practice operations, AI business case, quantifying AI business value, cost displacement vs revenue uplift vs retention vs premium monetisation, realisation rate, ROI sensitivity analysis, payback, break-even volume | `references/finops-ai-value-management.md` |
 | GenAI capacity planning, provisioned vs shared capacity, traffic shape, spillover, throughput units | `references/finops-genai-capacity.md` |
 | Self-hosted vs managed AI inference, build vs buy LLM, vLLM, SGLang, llama.cpp, GPU rental, RunPod, CoreWeave, Lambda, hidden cost surface, ML-Ops maturity rubric, hybrid routing (LiteLLM, Portkey) | `references/finops-ai-self-hosted-vs-managed.md` |
+| Open-weight model vendors on their own hosted APIs, DeepSeek, Qwen, Kimi, Moonshot, GLM, Z.ai, Chinese model APIs, open-weight pricing, time-based pricing, peak and off-peak token rates, vendor API vs third-party host channel choice, open-weight model licensing, GLM Coding Plan | `references/finops-open-weight-vendors.md` |
 | AWS billing data, CUR, Data Exports for FOCUS 1.2, Cost Explorer, EC2/compute rightsizing, SageMaker operational FinOps, cost allocation, governance, CloudFront flat-rate plans, S3 Files, multi-org billing, expensive IAM actions, cost-preventive SCPs, deny high-cost actions, sandbox account guardrails | `references/finops-aws.md` |
 | AWS Savings Plans, Reserved Instances, Spot, commitment decision tree, commitment portfolio liquidity, phased purchasing, EDP negotiation, Convertible RI exchange | `references/finops-aws-commitments.md` |
 | AWS per-service inefficiency catalogue, enumerated AWS optimisation patterns | `references/finops-aws-patterns.md` |

skills/cloud-finops/references/finops-ai-self-hosted-vs-managed.md

@@ -298,6 +298,8 @@ sources of bad recommendations from generic consultants. The OptimNow approach i
 ## References (other files in this skill)
 
 - `finops-for-ai.md` for AI cost mechanics, allocation, ROI framework
+- `finops-open-weight-vendors.md` for the third channel: buying an open-weight model
+  from the lab that trained it (DeepSeek, Qwen, Kimi, GLM hosted APIs)
 - `finops-genai-capacity.md` for provisioned vs shared capacity and traffic shape
 - `finops-anthropic.md`, `finops-bedrock.md`, `finops-azure-openai.md`, `finops-vertexai.md`
   for managed API specifics