Model Catalog
Providers, models, and per-user access control.
Current workspace
ProjectX Production
projectx-prod
● 6 providers · 142 models enabled
Catalog is workspace-scoped. Providers and aliases are per-workspace.
Active providers in projectx-prod
6 providers
connecting to 4 integrations · 184 total models exposed
Providers
6
1 degraded · 1 down
Integrations
4
Azure · OpenAI · Anthropic · DeepSeek
Total models
184
142 enabled · 24 disabled · 18 aliases
Last health check
42s ago
auto every 60s
| NAME | SLUG | INTEGRATION | ENDPOINT | MODELS EXPOSED | STATUS | CREATED BY | ACTIONS |
|---|---|---|---|---|---|---|---|
| azure-prod-eastus | az-prod-eastus | Azure Foundry | eastus.api.azure.com/openai/v1 | 82 | ● Healthy | jerry.du | Edit · Test |
| azure-prod-hk | az-prod-hk | Azure Foundry | acaiservices.services.ai.azure.com/anthropic/v1 | 14 (Anthropic only) | ● Healthy | jerry.du | Edit · Test |
| openai-direct | oai-direct | OpenAI | api.openai.com/v1 | 24 | ● Healthy | tony.wong | Edit · Test |
| anthropic-direct | anth-direct | Anthropic | api.anthropic.com/v1 | 8 | ⚠ Degraded 3 timeouts / 1h | amy.chen | Edit · Test |
| deepseek-cn | ds-cn | DeepSeek | api.deepseek.com/v1 | 6 | ● Healthy | jerry.du | Edit · Test |
| acai-prod | acai-prod | AcaiServices | acaiservices.services.ai.azure.com/openai | 18 | ● Down auth failed 5m ago | li.wei | Edit · Test |
Concept: Provider = workspace-scoped instance of an Integration with its own slug and credentials.
Call syntax:
@{provider_slug}/{model_name}. Example: @az-prod-eastus/claude-opus-4-7.
Aliases can shorten this — see Models tab.
Showing 142 of 184 models available in projectx-prod · 24 disabled · 18 aliases
| MODEL | FAMILY | CAPABILITY | PROVIDER(S) | ALIAS | CONTEXT | PRICING (in / out per 1M) | STATUS |
|---|---|---|---|---|---|---|---|
| claude-opus-4-7 | Anthropic | ChatReasoning | az-prod-hk, anth-direct (2) | @claude-opus-4-7-azure-hk | 200K | $15 / $75 | ● Enabled |
| claude-opus-4-8 | Anthropic | ChatReasoning | az-prod-hk (1) | — | 200K | $18 / $90 | ● Enabled |
| claude-sonnet-4-6 | Anthropic | Chat | az-prod-hk, anth-direct (2) | — | 200K | $3 / $15 | ● Enabled |
| claude-haiku-4-5-20251001 | Anthropic | Chat | az-prod-hk (1) | — | 200K | $0.80 / $4 | ● Enabled |
| gpt-4.1 | OpenAI | ChatVision | oai-direct, az-prod-eastus, acai-prod (3) | — | 1M | $5 / $20 | ● Enabled |
| gpt-4.1-mini | OpenAI | Chat | oai-direct, az-prod-eastus (2) | — | 1M | $0.40 / $1.60 | ● Enabled |
| gpt-4.1-nano | OpenAI | Chat | oai-direct, az-prod-eastus (2) | — | 1M | $0.10 / $0.40 | ● Enabled |
| gpt-4o | OpenAI | ChatVision | oai-direct, az-prod-eastus, acai-prod (3) | — | 128K | $2.50 / $10 | ● Enabled |
| gpt-4o-mini | OpenAI | Chat | oai-direct (1) | — | 128K | $0.15 / $0.60 | ● Enabled |
| o1-preview | OpenAI | Reasoning | oai-direct (1) | — | 128K | $15 / $60 | ⛔ Disabled |
| o1-mini | OpenAI | Reasoning | oai-direct (1) | — | 128K | $3 / $12 | ● Enabled |
| text-embedding-3-large | OpenAI | Embedding | oai-direct (1) | — | 8K | $0.13 / — | ● Enabled |
| text-embedding-3-small | OpenAI | Embedding | oai-direct (1) | — | 8K | $0.02 / — | ● Enabled |
| DeepSeek-V4-Pro | DeepSeek | ChatReasoning | ds-cn (1) | @deepseek-pro | 128K | $1.20 / $4.80 | ● Enabled |
| DeepSeek-V4-Flash | DeepSeek | Chat | ds-cn (1) | @deepseek-flash | 128K | $0.20 / $0.80 | ● Enabled |
| DeepSeek-R1 | DeepSeek | Reasoning | ds-cn, az-prod-eastus (2) | — | 64K | $0.55 / $2.20 | ● Enabled |
| Llama-4-Maverick-17B-128E-Instruct-FP8 | Llama | Chat | az-prod-eastus (1) | — | 128K | $0.18 / $0.72 | ● Enabled |
| Llama-3.3-70B-Instruct | Llama | Chat | az-prod-eastus (1) | — | 128K | $0.40 / $1.20 | ● Enabled |
| Mistral-Large-3 | Mistral | Chat | az-prod-eastus (1) | — | 128K | $2 / $6 | ● Enabled |
| Phi-4-reasoning | Phi | Reasoning | az-prod-eastus (1) | — | 16K | $0.05 / $0.20 | ● Enabled |
| Stable-Image-Ultra | Stable Diffusion | Image | az-prod-eastus (1) | — | — | $0.08 / image | ● Enabled |
| Kimi-K2.6 | Kimi | ChatReasoning | az-prod-eastus (1) | — | 128K | $1 / $4 | ⛔ Disabled |
Aliases let you create stable workspace-local names that hide provider routing. Example:
@deepseek-pro resolves to @ds-cn/DeepSeek-V4-Pro — change the alias target without breaking client code.
Disabling a model removes it from all configs and team whitelists; existing requests fail with 404 model_disabled.
Per-user model whitelist.
Control which users and teams can access which models in projectx-prod.
Default: users see only models granted via team membership. Platform Admins always see all.
Teams in this workspace
Select a team to edit its model whitelist
- ProjectX Backend12 members
- ProjectX Frontend8 members
- ProjectX QA5 members
- ProjectX Research6 members
- ProjectX DevOps4 members
ProjectX Research · Model Access
6 members can use the following models. Uncheck to remove access.
8 models currently granted · 14 models available in catalog
⚠ 5 unsaved changes
▶ Anthropic 4 / 4 selected all granted
▶ OpenAI 2 / 5 selected partial
▶ DeepSeek 1 / 3 selected partial
▶ Embedding 1 / 2 selected partial
▶ Llama 0 / 2 selected none
▶ Mistral 0 / 1 selected none
▶ Phi 0 / 1 selected none
▶ Image 0 / 1 selected none
User overrides
Grant or deny specific models to individual users, overriding team defaults.
| USER | OVERRIDE | RULE |
|---|---|---|
| amy.chen Research |
Granted | + claude-opus-4-8 extra |
| tony.wong Research |
Denied | − DeepSeek-V4-Pro sensitive workload |
| sarah.li Research |
Granted | + o1-mini reasoning eval |
📋 All access changes are audited.
Last change: jerry.du removed
gpt-4.1-mini from ProjectX Research · 2h ago.
View in Audit Log.