Update proxy for modern model APIs

This commit is contained in:
Your Name
2026-04-06 03:59:37 -07:00
parent 824adfbbb2
commit 8662eadea7
48 changed files with 1294 additions and 214 deletions
+8 -3
View File
@@ -45,11 +45,16 @@ You can also request Claude Instant, but support for this isn't fully implemente
### Supported model IDs
Users can send these model IDs to the proxy to invoke the corresponding models.
- **Claude**
- `anthropic.claude-v1` (~18k context, claude 1.3 -- EOL 2024-02-28)
- `anthropic.claude-v2` (~100k context, claude 2.0)
- `anthropic.claude-v2:1` (~200k context, claude 2.1)
- **Claude Instant**
- `anthropic.claude-instant-v1` (~100k context, claude instant 1.2)
- `anthropic.claude-haiku-4-5-20251001-v1:0`
- `anthropic.claude-sonnet-4-5-20250929-v1:0`
- `anthropic.claude-opus-4-1-20250805-v1:0`
- `anthropic.claude-3-5-haiku-20241022-v1:0`
- `anthropic.claude-sonnet-4-20250514-v1:0`
- `anthropic.claude-opus-4-20250514-v1:0`
For OpenAI-compatible callers, the proxy will also remap newer Claude-style names such as `claude-sonnet-4-5-20250929`, `claude-haiku-4-5-20251001`, `claude-opus-4-1-20250805`, and `claude-3-5-haiku-20241022` to the corresponding Bedrock model IDs.
## Note regarding logging
+3 -1
View File
@@ -20,7 +20,9 @@ AZURE_CREDENTIALS=contoso-ml:gpt4-8k:0123456789abcdef0123456789abcdef,northwind-
Note that each Azure deployment is assigned a model when you create it in the Azure OpenAI Service portal. If you want to use a different model, you'll need to create a new deployment, and therefore a new key to be added to the AZURE_CREDENTIALS environment variable. Each credential only grants access to one model.
### Supported model IDs
Users can send normal OpenAI model IDs to the proxy to invoke the corresponding models. For the most part they work the same with Azure. GPT-3.5 Turbo has an ID of "gpt-35-turbo" because Azure doesn't allow periods in model names, but the proxy should automatically convert this to the correct ID.
Users can send normal OpenAI model IDs to the proxy to invoke the corresponding models. The proxy now understands newer Azure-backed OpenAI model families such as GPT-4o, GPT-4.1, GPT-5 / GPT-5.2, o-series reasoning models, and GPT Image deployments including `gpt-image-1.5`, plus the newer Responses API route at `/proxy/azure/openai/v1/responses`.
GPT-3.5 Turbo still has an Azure-specific ID of `gpt-35-turbo` because Azure doesn't allow periods in model names, but the proxy will automatically normalize that for you.
As noted above, you can only use model IDs for which a deployment has been created and added to the proxy.
+8 -4
View File
@@ -29,7 +29,11 @@ GCP_CREDENTIALS=my-first-project:xxx@yyy.com:us-east5:-----BEGIN PRIVATE KEY----
## Supported model IDs
Users can send these model IDs to the proxy to invoke the corresponding models.
- **Claude**
- `claude-3-haiku@20240307`
- `claude-3-sonnet@20240229`
- `claude-3-opus@20240229`
- `claude-3-5-sonnet@20240620`
- `claude-haiku-4-5@20251001`
- `claude-sonnet-4-5@20250929`
- `claude-opus-4-1@20250805`
- `claude-3-5-haiku@20241022`
- `claude-sonnet-4@20250514`
- `claude-opus-4@20250514`
For OpenAI-compatible callers, the proxy will also remap Claude-style aliases like `claude-sonnet-4-5-20250929`, `claude-haiku-4-5-20251001`, and `claude-opus-4-1-20250805` to the corresponding Vertex AI model IDs.