#830
Add GLM 5.2 provider coverage
Track missing provider coverage candidates for GLM 5.2.
Provider candidates to review: AkashML, Ambient, DigitalOcean, FriendliAI, io.net, Morph, NovitaAI, Parasail, Phala, SiliconFlow, StreamLake, Wafer, Weights and Biases.
Already present or already represented in the catalogue: AtlasCloud, Baseten, Cloudflare, DeepInfra, Fireworks, GMICloud, Together AI, Venice, Z.ai.
Tasks: verify each remaining candidate, add missing provider-model rows where reliable metadata is available, capture model IDs, pricing, context length and capability metadata, avoid duplicate provider rows, and update relevant tests or generated data.
Activity
2
opencode-agent[bot]4mo agoNow I have enough information to write the prompt.
Issue understanding
This is a data coverage task for GLM 5.2 (model `z-ai/glm-5.2`). The model definition already exists at `packages/data/catalog/src/data/models/z-ai/glm-5.2/model.json`. Several providers already have entries (AtlasCloud, Baseten, Cloudflare, Crofai, DeepInfra, Fireworks, GMICloud, Nebius Token Factory, Novita, Together, Venice, Venice E2EE, Z.ai). The remaining providers need to be reviewed and rows added if they support GLM 5.2. Each entry requires: a row in the provider's `models.json`, and a pricing file under `pricing//z-ai-glm-5.2/text.generate/pricing.json`.AI Agent Prompt (Copy/Paste)
This task is about adding GLM 5.2 provider coverage to the AI Stats data catalog.
The model `z-ai/glm-5.2` already has its model definition at `packages/data/catalog/src/data/models/z-ai/glm-5.2/model.json`.
Review the following providers for GLM 5.2 support:
AkashML, Ambient (no dir yet), AtlasCloud (already has glm-5.2), Baseten (already has), Cloudflare (already has), DeepInfra (already has), DigitalOcean (has glm-5 but NOT 5.2), Fireworks (already has), FriendliAI (has glm-5/5.1 but NOT 5.2), GMICloud (already has), io.net (no dir yet), Morph (no glm-5x entries), NovitaAI (already has), Parasail (no glm-5x entries), Phala (no glm-5x entries), SiliconFlow (has glm-5/5.1 but NOT 5.2), StreamLake (no dir yet), Together AI (already has), Venice (already has), Wafer (no dir yet), Weights and Biases (has glm-5/5.1 but NOT 5.2), Z.ai (already has).
For each provider:
1. Check if they actually offer GLM 5.2 (look at their API docs or existing entries for 5.1/5 patterns in `packages/data/catalog/src/data/api_providers/<provider>/models.json`).
2. If they do, add a model entry to their `models.json` using the established pattern (fields: api_model_id, provider_api_model_id, provider_model_slug, internal_model_id, is_active_gateway, quantization_scheme, input_modalities, output_modalities, context_length, max_output_tokens, effective_from, effective_to, capabilities).
3. Add a pricing file at `packages/data/catalog/src/data/pricing/<provider>/z-ai-glm-5.2/text.generate/pricing.json` (see `pricing/together/z-ai-glm-5.2/text.generate/pricing.json` for the canonical structure; Z.ai base pricing: $1.40/1M input, $0.26/1M cached, $4.40/1M output).
4. For providers with no directory yet (Ambient, io.net, StreamLake, Wafer), create the full provider directory with `api_provider.json` (model sibling dirs), `models.json` with the GLM 5.2 entry, and pricing file — or skip if they don't offer GLM models at all.
5. Run `pnpm validate:data` and `pnpm --filter @ai-stats/web test` to verify no regressions.
Inspect each provider's existing `models.json` to understand the exact model entry format before adding rows. Avoid duplicate rows.
AI-194