Repository navigation
feat(v2): complete model APIs, packages and prices for Azure, Bedrock, Databricks, Snowflake, Copilot and GitLab - #11136
Merged
Merged
Conversation
Contributor
Action items
|
Collaborator
Author
|
Confirmed: the full |
…for its v1 shape A leading [api.responses] on an @ai-sdk/openai or @ai-sdk/azure provider is the SDK's default and should publish no v1 shape, as a file with no tables does. Two Infer files that relied on the old reading keep their shape with legacy_shape. v1 is byte-identical.
createXai().languageModel() is the Responses model, so a leading [api.responses] on xAI is the SDK's default and publishes no v1 shape. No existing file is affected.
…r its OpenAI models
…ions (pending approval) Azure's Responses API lists only OpenAI models (and not gpt-3.5-turbo or gpt-4-turbo), but these 26 models had no API table, so they inherited @ai-sdk/azure's Responses default. Each now lists Chat Completions. v1: these 26 models (and their azure-cognitive-services aliases) gain provider.shape = "completions".
rekram1-node
force-pushed
the
imp-labs-2
branch
from
October 23, 2026 05:25
a7cd4ee to
3d14a31
Compare
Contributor
|
No actionable findings. |
github-actions
Bot
added
the
reviewer: ready
rekram1-node
enabled auto-merge (squash)
October 23, 2026 05:47
rekram1-node
disabled auto-merge
October 23, 2026 05:47
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Builds on #11135. Its first two commits are the same generator fix: a first API table with no
aisdktakes the provider'snpmfor its v1 shape, so Azure can list Responses then Chat Completions without changing v1. Merge #11135 first.Problem
Cloud and gateway files listed one API per model. Bedrock Mantle, Azure Chat Completions, Copilot's Claude Chat Completions, and the Databricks and Snowflake Messages and Responses endpoints were all missing. GitLab Claude models were labeled Chat Completions, even though they go through the Anthropic proxy. Azure Foundry models had no table at all, so they inherited
@ai-sdk/azure's Responses default, which Azure doesn't serve for them.Change
The first table is always what the model already published.
/openai/v1for GPT, Grok 4.3, Gemma 4;/v1for gpt-oss);[api.converse]+ Mantle Chat (/v1) on 26 open-weight models;[api.converse]+ Mantle Messages (/anthropic/v1) on 6 bare Claude IDsGET /v1/models, then each API checked live in us-east-1azure/chat) next to Responses on 26 OpenAI chat modelsGET /modelssupported_endpoints(recorded in provider.toml)[api.messages]on 14 Claude models (was a derived Chat Completions); Responses next to Chat on 17 OpenAI models/serving-endpoints/responses) on 12 GPT models; Messages (/serving-endpoints/anthropic/v1) on 8 Claude models/api/v2/cortex/v1/messages) on 10 Claude modelsPackages. Most specific existing entry:
amazon-bedrock/mantle/chat|responses|messages,azure/chat,openai-compatible/responses,anthropic-compatible. Prices: nothing missing that v2 can hold (Databricks batch is per DBU-hour and stays unmodeled).Seven Databricks Claude files and phi-4-multimodal were rewritten by the formatter: number underscores only (
8000->8_000).Pending user approval (last commit)
26 Azure models (Grok, Llama, Mistral, Phi, DeepSeek V3.2, Cohere Command A, gpt-3.5-turbo-0125/1106, gpt-4-turbo, gpt-4-turbo-vision) now list Chat Completions only. Azure's Responses list covers only OpenAI models, and not these legacy ones. v1: the 26 models and their
azure-cognitive-servicesaliases gainprovider.shape = "completions"(52 entries).Out of scope / report only
anthropic.claude-opus-4-8, etc.) fail Converse on demand ("on-demand throughput isn't supported"). Converse stays first for provisioned use. Making Mantle Messages first would change v1.messageskey with Mantle and has no OpenCode entry. Databricks Open Responses lives at/serving-endpoints/open-responses(no/responsessuffix), so no SDK can reach it from a base URL. Azuregpt-3.5-turbo-instructis legacy Completions only (no protocol for it).supported_endpointslists no Chat Completions for them.Verification
bun validate: passes.bun run v1:check: byte-identical without the pending commit; with it, only the 52provideradditions above.bun test: 709 pass, 10 fail (the same 10 asdev).dev: API changes only. amazon-bedrock 47 models, azure 52 (26 first-API changes are the pending commit), github-copilot 11, gitlab 31, databricks 20, snowflake-cortex 10.