chore: openai reasoning model cleanup by lifeizhou-ap · Pull Request #7529 · block/goose

lifeizhou-ap · 2026-02-26T03:01:09Z

Summary

removed deprecate max_token request param in openai compatible api
use canonical model reasoning data to check whether model is reasoning model

Type of Change

Testing

Unit test and manual testing

…d also use canonical model reasoning data to check whether model is reasoning model

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 61efebd998

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

chatgpt-codex-connector · 2026-02-26T03:07:14Z

crates/goose/src/providers/formats/openai.rs

+    payload.as_object_mut().unwrap().insert(
+        "max_completion_tokens".to_string(),
+        json!(model_config.max_output_tokens()),


Restore max_tokens for non-reasoning chat completions

This now unconditionally writes max_completion_tokens, but this formatter is shared by multiple non-OpenAI providers (for example openai_compatible, ollama, tetrate, githubcopilot, litellm, and openrouter) that still post to /chat/completions schemas expecting max_tokens; with this change, non-reasoning requests no longer include max_tokens at all, which can cause 400s or silently drop output limits on those backends. Keep max_completion_tokens for reasoning models, but preserve a max_tokens path for standard chat-completions models.

Useful? React with 👍 / 👎.

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 2bb0e8e9d1

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

chatgpt-codex-connector · 2026-02-26T05:17:56Z

crates/goose/src/providers/formats/databricks.rs

+        payload.as_object_mut().unwrap().insert(
+            "max_completion_tokens".to_string(),
+            json!(model_config.max_output_tokens()),


Preserve max_tokens for non-reasoning Databricks requests

create_request now always emits max_completion_tokens in the normal (non-CLAUDE_THINKING_ENABLED) path, so non-reasoning Databricks models no longer receive max_tokens. This path is used for standard Databricks chat invocations (for example llama/claude endpoints), and switching the parameter can cause provider-side 400s or ignored output limits on backends that still implement chat-completions semantics with max_tokens; keep max_completion_tokens only for reasoning-model payloads.

Useful? React with 👍 / 👎.

michaelneale · 2026-02-26T06:05:39Z

crates/goose/src/model.rs

    }

+    pub fn is_openai_reasoning_model(&self) -> bool {
+        const PROVIDER_PREFIXES: &[&str] = &["goose-", "databricks-"];


provider or model name? as that seems odd?

should be DATABRICKS_MODEL_NAME_PREFIXES

michaelneale

much nicer but provider/model name prefix is a little confusing for me.

lifeizhou-ap · 2026-02-26T06:20:41Z

much nicer but provider/model name prefix is a little confusing for me.

yes @michaelneale! I was a bit confused initially too. Looking into the code, I found the databricks provider currently is much for block databricks which we know the convention of our model names. If we want to have a general purpose databricks provider, we have to have a common schema for returning the servingpoints.

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: ace6b450c0

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

chatgpt-codex-connector · 2026-02-26T06:52:43Z

crates/goose/src/model.rs


+    pub fn is_openai_reasoning_model(&self) -> bool {
+        const DATABRICKS_MODEL_NAME_PREFIXES: &[&str] = &["goose-", "databricks-"];
+        const REASONING_PREFIXES: &[&str] = &["o1", "o3", "o4", "gpt-5"];


Keep o2 models classified as reasoning

is_openai_reasoning_model no longer treats o2* names as reasoning models because REASONING_PREFIXES dropped "o2". This helper now drives request shaping in both OpenAI and Databricks formatters, so o2 requests will be sent down the non-reasoning path (including temperature and without reasoning-specific handling), which can trigger 400s or behavior mismatches on endpoints that still expose o2 under o-series semantics.

Useful? React with 👍 / 👎.

* main: feat: add goose-powered release notes generator workflow (#7503)

* main: (46 commits) chore(deps): bump minimatch from 10.1.1 to 10.2.3 in /evals/open-model-gym/suite (#7498) chore(deps): bump swiper from 11.2.10 to 12.1.2 in /documentation (#7368) Better network failure error & antrhopic retry (#7595) feat: make the text bar persistent and add a queue for messages (#7560) fix: outdated clippy command in goosehints (#7590) chore(deps): bump hono from 4.11.7 to 4.12.1 in /evals/open-model-gym/mcp-harness (#7417) chore(deps-dev): bump ajv from 6.12.6 to 6.14.0 in /ui/desktop (#7437) chore(deps): bump ajv from 8.17.1 to 8.18.0 in /evals/open-model-gym/mcp-harness (#7491) chore(deps): bump hono from 4.12.0 to 4.12.2 in /ui/desktop (#7515) chore(deps-dev): bump rollup from 4.57.1 to 4.59.0 in /ui/desktop (#7522) chore(deps): bump minimatch in /ui/desktop (#7572) fix: validate configure probe for streaming providers (#7564) Dockerfile: add missing build/runtime dependencies (#7546) fix(claude-code): Permission routing for smart-approve (#7501) Add base_path field to custom provider config (#7558) fix(cli): avoid debug logging by default in CLI (#7569) fix: panic on corrupted permission.yaml instead of silently allowing all (#7432) (#7458) fix(openai): handle null reasoning effort in Responses API (#7469) Allow GOOSE_NODE_DIR override in batch file (#7422) feat: add analyze platform extension with tree-sitter AST parsing (#7542) ...

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 84e1e20cd6

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

chatgpt-codex-connector · 2026-03-02T14:59:49Z

crates/goose/src/providers/databricks.rs

 pub const DATABRICKS_KNOWN_MODELS: &[&str] = &[
    "databricks-claude-sonnet-4-5",
-    "databricks-claude-3-7-sonnet",
    "databricks-meta-llama-3-3-70b-instruct",
    "databricks-meta-llama-3-1-405b-instruct",


Restore removed Databricks models in known model metadata

DATABRICKS_KNOWN_MODELS now omits databricks-claude-3-7-sonnet and databricks-dbrx-instruct, but this constant is still passed into ProviderMetadata::new(...) for the Databricks provider and is used as the static/fallback model catalog exposed to clients. Removing these entries makes still-referenced Databricks models undiscoverable in provider metadata (including UI selection/fallback flows when dynamic model fetch is unavailable), even though the codebase still treats them as valid model names in other places.

Useful? React with 👍 / 👎.

* origin/main: (27 commits) feat: return structured {stdout, stderr} from shell tool with output schema (#7604) Improve custom provider creation experience (#7541) fix(scheduler): schedules added via CLI showing up in UI (#7594) chore: openai reasoning model cleanup (#7529) chore(deps): bump hono from 4.12.1 to 4.12.3 in /evals/open-model-gym/mcp-harness (#7585) chore(deps): bump minimatch from 10.1.1 to 10.2.3 in /evals/open-model-gym/suite (#7498) chore(deps): bump swiper from 11.2.10 to 12.1.2 in /documentation (#7368) Better network failure error & antrhopic retry (#7595) feat: make the text bar persistent and add a queue for messages (#7560) fix: outdated clippy command in goosehints (#7590) chore(deps): bump hono from 4.11.7 to 4.12.1 in /evals/open-model-gym/mcp-harness (#7417) chore(deps-dev): bump ajv from 6.12.6 to 6.14.0 in /ui/desktop (#7437) chore(deps): bump ajv from 8.17.1 to 8.18.0 in /evals/open-model-gym/mcp-harness (#7491) chore(deps): bump hono from 4.12.0 to 4.12.2 in /ui/desktop (#7515) chore(deps-dev): bump rollup from 4.57.1 to 4.59.0 in /ui/desktop (#7522) chore(deps): bump minimatch in /ui/desktop (#7572) fix: validate configure probe for streaming providers (#7564) Dockerfile: add missing build/runtime dependencies (#7546) fix(claude-code): Permission routing for smart-approve (#7501) Add base_path field to custom provider config (#7558) ...

* main: (74 commits) feat: return structured {stdout, stderr} from shell tool with output schema (#7604) Improve custom provider creation experience (#7541) fix(scheduler): schedules added via CLI showing up in UI (#7594) chore: openai reasoning model cleanup (#7529) chore(deps): bump hono from 4.12.1 to 4.12.3 in /evals/open-model-gym/mcp-harness (#7585) chore(deps): bump minimatch from 10.1.1 to 10.2.3 in /evals/open-model-gym/suite (#7498) chore(deps): bump swiper from 11.2.10 to 12.1.2 in /documentation (#7368) Better network failure error & antrhopic retry (#7595) feat: make the text bar persistent and add a queue for messages (#7560) fix: outdated clippy command in goosehints (#7590) chore(deps): bump hono from 4.11.7 to 4.12.1 in /evals/open-model-gym/mcp-harness (#7417) chore(deps-dev): bump ajv from 6.12.6 to 6.14.0 in /ui/desktop (#7437) chore(deps): bump ajv from 8.17.1 to 8.18.0 in /evals/open-model-gym/mcp-harness (#7491) chore(deps): bump hono from 4.12.0 to 4.12.2 in /ui/desktop (#7515) chore(deps-dev): bump rollup from 4.57.1 to 4.59.0 in /ui/desktop (#7522) chore(deps): bump minimatch in /ui/desktop (#7572) fix: validate configure probe for streaming providers (#7564) Dockerfile: add missing build/runtime dependencies (#7546) fix(claude-code): Permission routing for smart-approve (#7501) Add base_path field to custom provider config (#7558) ...

* origin/main: feat: return structured {stdout, stderr} from shell tool with output schema (#7604) Improve custom provider creation experience (#7541) fix(scheduler): schedules added via CLI showing up in UI (#7594) chore: openai reasoning model cleanup (#7529) chore(deps): bump hono from 4.12.1 to 4.12.3 in /evals/open-model-gym/mcp-harness (#7585) chore(deps): bump minimatch from 10.1.1 to 10.2.3 in /evals/open-model-gym/suite (#7498) chore(deps): bump swiper from 11.2.10 to 12.1.2 in /documentation (#7368) Better network failure error & antrhopic retry (#7595) feat: make the text bar persistent and add a queue for messages (#7560)

* origin/main: (107 commits) Merge platform/builtin extensions (#7630) Clean up stale references to removed components (#7644) fix: scope empty session reuse to current window to prevent session mixing (#7602) fix: prevent abort in local inference (#7633) Revert git patch for llama-cpp-2 (#7642) docs: update recipe usage step to reflect auto-submit behavior (#7639) docs: add guide for customizing the sidebar (#7638) docs: update Claude Code approve behavior and model list in cli-providers guide (#7448) fix: restore provider and extensions for LRU-evicted sessions (#7616) Restore goosed logging (#7622) feat: return structured {stdout, stderr} from shell tool with output schema (#7604) Improve custom provider creation experience (#7541) fix(scheduler): schedules added via CLI showing up in UI (#7594) chore: openai reasoning model cleanup (#7529) chore(deps): bump hono from 4.12.1 to 4.12.3 in /evals/open-model-gym/mcp-harness (#7585) chore(deps): bump minimatch from 10.1.1 to 10.2.3 in /evals/open-model-gym/suite (#7498) chore(deps): bump swiper from 11.2.10 to 12.1.2 in /documentation (#7368) Better network failure error & antrhopic retry (#7595) feat: make the text bar persistent and add a queue for messages (#7560) fix: outdated clippy command in goosehints (#7590) ... # Conflicts: # Cargo.lock # Cargo.toml # crates/goose-server/src/commands/agent.rs # crates/goose-server/src/main.rs # crates/goose-server/src/routes/reply.rs

removed deprecate max_token request param in openai compatible api an…

61efebd

…d also use canonical model reasoning data to check whether model is reasoning model

chatgpt-codex-connector bot reviewed Feb 26, 2026

View reviewed changes

schema update

2bb0e8e

lifeizhou-ap requested review from DOsinga and katzdave February 26, 2026 05:13

chatgpt-codex-connector bot reviewed Feb 26, 2026

View reviewed changes

michaelneale reviewed Feb 26, 2026

View reviewed changes

michaelneale approved these changes Feb 26, 2026

View reviewed changes

updated the variable name

ace6b45

chatgpt-codex-connector bot reviewed Feb 26, 2026

View reviewed changes

katzdave approved these changes Feb 26, 2026

View reviewed changes

lifeizhou-ap added 4 commits March 2, 2026 09:29

Merge branch 'main' into lifei/clean-deprecate-parameter

25e17b2

* main: feat: add goose-powered release notes generator workflow (#7503)

fixed merge

7debbd3

fixed fmt

84e1e20

chatgpt-codex-connector bot reviewed Mar 2, 2026

View reviewed changes

lifeizhou-ap added this pull request to the merge queue Mar 2, 2026

Merged via the queue into main with commit 82facd4 Mar 2, 2026
27 of 28 checks passed

lifeizhou-ap deleted the lifei/clean-deprecate-parameter branch March 2, 2026 15:47

github-actions bot mentioned this pull request Mar 2, 2026

chore(release): release version 1.27.0 (minor) #7611

Merged

craigwalkeruk pushed a commit to craigwalkeruk/custom-goose that referenced this pull request Mar 5, 2026

chore: openai reasoning model cleanup (block#7529)

16fc93b

lifeizhou-ap mentioned this pull request Mar 13, 2026

Deprecate "max_tokens" in openai provider api request payload #5331

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

chore: openai reasoning model cleanup#7529

chore: openai reasoning model cleanup#7529
lifeizhou-ap merged 7 commits intomainfrom
lifei/clean-deprecate-parameter

lifeizhou-ap commented Feb 26, 2026

Uh oh!

chatgpt-codex-connector bot left a comment

Uh oh!

chatgpt-codex-connector bot Feb 26, 2026

Uh oh!

chatgpt-codex-connector bot left a comment

Uh oh!

chatgpt-codex-connector bot Feb 26, 2026

Uh oh!

michaelneale Feb 26, 2026

Uh oh!

lifeizhou-ap Feb 26, 2026 •

edited

Loading

Uh oh!

michaelneale left a comment

Uh oh!

lifeizhou-ap commented Feb 26, 2026 •

edited

Loading

Uh oh!

chatgpt-codex-connector bot left a comment

Uh oh!

chatgpt-codex-connector bot Feb 26, 2026

Uh oh!

chatgpt-codex-connector bot left a comment

Uh oh!

chatgpt-codex-connector bot Mar 2, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Conversation

lifeizhou-ap commented Feb 26, 2026

Summary

Type of Change

Testing

Uh oh!

chatgpt-codex-connector bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

chatgpt-codex-connector bot Feb 26, 2026

Choose a reason for hiding this comment

Uh oh!

chatgpt-codex-connector bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

chatgpt-codex-connector bot Feb 26, 2026

Choose a reason for hiding this comment

Uh oh!

michaelneale Feb 26, 2026

Choose a reason for hiding this comment

Uh oh!

lifeizhou-ap Feb 26, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

michaelneale left a comment

Choose a reason for hiding this comment

Uh oh!

lifeizhou-ap commented Feb 26, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

chatgpt-codex-connector bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

chatgpt-codex-connector bot Feb 26, 2026

Choose a reason for hiding this comment

Uh oh!

chatgpt-codex-connector bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

chatgpt-codex-connector bot Mar 2, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

lifeizhou-ap Feb 26, 2026 •

edited

Loading

lifeizhou-ap commented Feb 26, 2026 •

edited

Loading