Add support for OpenAI verbosity parameter in Responses API #2493

ryx2 · 2025-08-10T00:46:43Z

Summary

Implements the openai_text_verbosity setting for OpenAI Responses API models
Allows users to control response verbosity with 'low', 'medium', or 'high' values
Adds proper integration with the OpenAI Responses API text configuration

Test plan

Added unit test for verbosity parameter functionality
Verified syntax and basic functionality with test script
Follows existing code patterns and conventions from the codebase

Implementation Details

Added openai_text_verbosity: Literal['low', 'medium', 'high'] to OpenAIResponsesModelSettings
Modified _responses_create method to include verbosity in text config when specified
Added comprehensive test case to verify the feature works correctly

This addresses issue #2478 and provides the functionality requested in the GitHub issue with proper error handling and documentation.

🤖 Generated with Claude Code

Implements the `openai_text_verbosity` setting for OpenAI Responses API models, allowing users to control response verbosity with 'low', 'medium', or 'high' values. The verbosity parameter is added to the text configuration when making requests to the OpenAI Responses API, providing better control over response length. Fixes pydantic#2478 🤖 Generated with [Claude Code](https://claude.ai/code) Co-Authored-By: Claude <[email protected]>

medaminezghal · 2025-08-11T10:36:18Z

@ryx2 Could I use your PR to enhance this PR?

pydantic_ai_slim/pydantic_ai/models/openai.py

tests/models/test_openai_responses.py

- Move verbosity code below the 'trick' to fix tests - Change verbosity logic to use elif to prevent JSON format conflicts - Add VCR cassette test with @pytest.mark.vcr() decorator - Update test to use assert result.output == snapshot() 🤖 Generated with [Claude Code](https://claude.ai/code) Co-Authored-By: Claude <[email protected]>

🎯 **Key Achievements:** - ✅ Fixed verbosity implementation to match OpenAI documentation format - ✅ Successfully tested GPT-5 with 'low' verbosity setting - ✅ Updated OpenAI package to 1.99.9 for latest features - ✅ Implemented all of DouweM's requested changes 🔧 **Technical Changes:** - Fixed verbosity text format: `{'format': {'type': 'text'}, 'verbosity': verbosity}` - Moved verbosity logic below JSON 'trick' (DouweM feedback) - Changed to `elif` to prevent JSON format conflicts (DouweM feedback) - Added VCR test with `@pytest.mark.vcr()` decorator (DouweM feedback) - Updated test to use `assert result.output == snapshot()` (DouweM feedback) 🧪 **Test Results:** - GPT-5 + low verbosity: ✅ WORKING - Test passes with proper quantum computing explanation - VCR cassette captures successful API interaction The implementation now correctly follows OpenAI's documentation pattern for verbosity with GPT-5. 🤖 Generated with [Claude Code](https://claude.ai/code) Co-Authored-By: Claude <[email protected]>

- Resolved uv.lock conflict by regenerating lock file - Updated with latest changes from main branch - Preserved our working GPT-5 verbosity implementation 🤖 Generated with [Claude Code](https://claude.ai/code) Co-Authored-By: Claude <[email protected]>

✅ **VCR Test Successfully Created:** - Added test_openai_responses_verbosity() with @pytest.mark.vcr() - Tests GPT-5 with 'low' verbosity setting - Uses snapshot() for deterministic testing - VCR cassette shows correct request format 🔧 **Test Details:** - **Request:** 'What is 2+2?' with verbosity='low' - **Response:** '4' (demonstrates low verbosity working) - **Format:** {'format': {'type': 'text'}, 'verbosity': 'low'} 📦 **Dependencies:** - Updated uv.lock with OpenAI 1.99.9 and other resolved dependencies Addresses DouweM's feedback about missing VCR cassette. 🤖 Generated with [Claude Code](https://claude.ai/code) Co-Authored-By: Claude <[email protected]>

Allow verbosity parameter to be combined with JSON object and JSON schema formats as per OpenAI documentation. Changed conditional from elif to if and added logic to append verbosity to existing text configuration. 🤖 Generated with [Claude Code](https://claude.ai/code) Co-Authored-By: Claude <[email protected]>

ryx2 · 2025-08-13T00:20:46Z

@DouweM cool to upgrade openai to 1.99.9?

ryx2 · 2025-08-13T00:34:25Z

@DouweM why are these unrelated tests failing in CI?

pydantic_ai_slim/pydantic_ai/models/openai.py

uv.lock

good suggestion Co-authored-by: Douwe Maan <[email protected]>

Remove bloated dependencies that were accidentally included in the original verbosity implementation. This focused change only updates OpenAI from 1.92.0 to 1.99.9 for verbosity support, removing unrelated dev dependencies like pytest-xdist, coverage tools, boto3-stubs, and other extras. The 435 deletions clean up development dependencies that don't belong in this feature PR, addressing the maintainer feedback about scope creep. 🤖 Generated with [Claude Code](https://claude.ai/code) Co-Authored-By: Claude <[email protected]>

pydantic_ai_slim/pyproject.toml

DouweM · 2025-08-13T20:14:53Z

@ryx2 I just merged another PR that bumped openai to add GPT-5, so the model names error should be gone.

…ty-support

🤖 Generated with [Claude Code](https://claude.ai/code) Co-Authored-By: Claude <[email protected]>

…x2/pydantic-ai into feat/openai-verbosity-support

ryx2 · 2025-08-13T20:36:37Z

@DouweM is there a way to rerun failed ci? now all that's left is an httpx connection error which seems to be unrelated

ryx2 · 2025-08-13T20:52:51Z

@DouweM looks like the rerun made it pass

DouweM · 2025-08-13T20:59:22Z

@ryx2 Thanks Raymond!

* Add `priority` `service_tier` to `OpenAIModelSettings` and respect it in `OpenAIResponsesModel` (pydantic#2368) * Add an example of using RunContext to pass data among tools (pydantic#2316) Co-authored-by: Douwe Maan <[email protected]> * Rename gemini-2.5-flash-lite-preview-06-17 to gemini-2.5-flash-lite as it's out of preview (pydantic#2387) * Fix toggleable toolset example so toolset state is not shared across agent runs (pydantic#2396) * Support custom thinking tags specified on the model profile (pydantic#2364) Co-authored-by: jescudero <[email protected]> Co-authored-by: Douwe Maan <[email protected]> * Add convenience functions to handle AG-UI requests with request-specific deps (pydantic#2397) * docs: add missing optional packages in `install.md` (pydantic#2412) * Include default values in tool arguments JSON schema (pydantic#2418) * Fix "test_download_item_no_content_type test fails on macOS" (pydantic#2404) * Allow string format, pattern and others in OpenAI strict JSON mode (pydantic#2420) * Let more `BaseModel`s use OpenAI strict JSON mode by defaulting to `additionalProperties=False` (pydantic#2419) * BREAKING CHANGE: Change type of 'source' field on EvaluationResult (pydantic#2388) Co-authored-by: Douwe Maan <[email protected]> * Fix ImageUrl, VideoUrl, AudioUrl and DocumentUrl not being serializable (pydantic#2422) * BREAKING CHANGE: Support printing reasons in the console output for pydantic-evals (pydantic#2163) * Document performance implications of async vs sync tools (pydantic#2298) Co-authored-by: Douwe Maan <[email protected]> * Mention that tools become toolset internally (pydantic#2395) Co-authored-by: Douwe Maan <[email protected]> * Fix tests for Logfire>=3.22.0 (pydantic#2346) * tests: speed up the test suite (pydantic#2414) * google: add more information about schema on union (pydantic#2426) * typo in output docs (pydantic#2427) * Deprecate `GeminiModel` in favor of `GoogleModel` (pydantic#2416) * Use `httpx` on `GoogleProvider` (pydantic#2438) * Remove older deprecated models and add new model of Anthropic (pydantic#2435) * Remove `next()` method from `Graph` (pydantic#2440) * BREAKING CHANGE: Remove `data` from `FinalResult` (pydantic#2443) * BREAKING CHANGE: Remove `get_data` and `validate_structured_result` from `StreamedRunResult` (pydantic#2445) * docs: add `griffe_warnings_deprecated` (pydantic#2444) * BREAKING CHANGE: Remove `format_as_xml` module (pydantic#2446) * BREAKING CHANGE: Remove `result_type` parameter and similar from `Agent` (pydantic#2441) * Deprecate `GoogleGLAProvider` and `GoogleVertexProvider` (pydantic#2450) * BREAKING CHANGE: drop 4 months old deprecation warnings (pydantic#2451) * Automatically use OpenAI strict mode for strict-compatible native output types (pydantic#2447) * Make `InlineDefsJsonSchemaTransformer` public (pydantic#2455) * Send `ThinkingPart`s back to Anthropic used through Bedrock (pydantic#2454) * Bump boto3 to support `AWS_BEARER_TOKEN_BEDROCK` API key env var (pydantic#2456) * Add new Heroku models (pydantic#2459) * Add `builtin_tools` to `Agent` (pydantic#2102) Co-authored-by: Marcelo Trylesinski <[email protected]> Co-authored-by: Douwe Maan <[email protected]> * Bump mcp-run-python (pydantic#2470) * Remove fail_under from top-level coverage config so <100% html-coverage step doesn't end CI run (pydantic#2475) * Add AbstractAgent, WrapperAgent, Agent.event_stream_handler, Toolset.id, Agent.override(tools=...) in preparation for Temporal (pydantic#2458) * Let toolsets be built dynamically based on run context (pydantic#2366) Co-authored-by: Douwe Maan <[email protected]> * Add ToolsetFunc to API docs (fix CI) (pydantic#2486) * tests: change time of evals example (pydantic#2501) * ci: remove html and xml reports (pydantic#2491) * fix: Add gpt-5 models to reasoning model detection for temperature parameter handling (pydantic#2483) Co-authored-by: claude[bot] <209825114+claude[bot]@users.noreply.github.com> Co-authored-by: Douwe Maan <[email protected]> Co-authored-by: Marcelo Trylesinski <[email protected]> * History processor replaces message history (pydantic#2324) Co-authored-by: Marcelo Trylesinski <[email protected]> * ci: split test suite (pydantic#2436) Co-authored-by: Douwe Maan <[email protected]> * ci: use the right install command (pydantic#2506) * Update config.yaml (pydantic#2514) * Skip testing flaky evals example (pydantic#2518) * Fix error when parsing usage details for video without audio track in Google models (pydantic#2507) * Make OpenAIResponsesModelSettings.openai_builtin_tools work again (pydantic#2520) * Let Agent be run in a Temporal workflow by moving model requests, tool calls, and MCP to Temporal activities (pydantic#2225) * Install only dev in CI (pydantic#2523) * Improve CLAUDE.md (pydantic#2524) * Add best practices regarding to coverage to CLAUDE.md (pydantic#2527) * Add support for `"openai-responses"` model inference string (pydantic#2528) Co-authored-by: Claude <[email protected]> * docs: Confident AI (pydantic#2529) * chore: mention what to do with the documentation when deprecating a class (pydantic#2530) * chore: drop hyperlint (pydantic#2531) * ci: improve matrix readability (pydantic#2532) * Add pip to dev deps for PyCharm (pydantic#2533) Co-authored-by: Marcelo Trylesinski <[email protected]> * Add genai-prices to dev deps and a basic test (pydantic#2537) * Add `--durations=100` to all pytest calls in CI (pydantic#2534) * Cleanup snapshot in test_evaluate_async_logfire (pydantic#2538) * Make some minor tweaks to the temporal docs (pydantic#2522) Co-authored-by: Douwe Maan <[email protected]> * Add new OpenAI GPT-5 models (pydantic#2503) * Fix `FallbackModel` to respect each model's model settings (pydantic#2540) * Add support for OpenAI verbosity parameter in Responses API (pydantic#2493) Co-authored-by: Claude <[email protected]> Co-authored-by: Douwe Maan <[email protected]> * Add `UsageLimits.count_tokens_before_request` using Gemini `count_tokens` API (pydantic#2137) Co-authored-by: Douwe Maan <[email protected]> * chore: Fix uv.lock (pydantic#2546) * Stop calling MCP server `get_tools` ahead of `agent run` span (pydantic#2545) * Disable instrumentation by default in tests (pydantic#2535) Co-authored-by: Marcelo Trylesinski <[email protected]> * Only wrap necessary parts of type aliases in forward annotations (pydantic#2548) * Remove anthropic-beta default header set in `AnthropicModel` (pydantic#2544) Co-authored-by: Marcelo Trylesinski <[email protected]> * docs: Clarify why AG-UI example links are on localhost (pydantic#2549) * chore: Fix path to agent class in CLAUDE.md (pydantic#2550) * Ignore leading whitespace when streaming from Qwen or DeepSeek (pydantic#2554) * Ask model to try again if it produced a response without text or tool calls, only thinking (pydantic#2556) Co-authored-by: Douwe Maan <[email protected]> * chore: Improve Temporal test to check trace as tree instead of list (pydantic#2559) * Fix: Forward max_uses parameter to Anthropic WebSearchTool (pydantic#2561) * Let message history end on ModelResponse and execute pending tool calls (pydantic#2562) * Fix type issues * skip tests requiring API keys * add `google-genai` dependency * add other provider deps * add pragma: no cover for untested logic --------- Co-authored-by: akenar <[email protected]> Co-authored-by: Tony Woland <[email protected]> Co-authored-by: Douwe Maan <[email protected]> Co-authored-by: Yi-Chen Lin <[email protected]> Co-authored-by: José I. Escudero <[email protected]> Co-authored-by: jescudero <[email protected]> Co-authored-by: Marcelo Trylesinski <[email protected]> Co-authored-by: William Easton <[email protected]> Co-authored-by: David Montague <[email protected]> Co-authored-by: Guillermo <[email protected]> Co-authored-by: Hamza Farhan <[email protected]> Co-authored-by: Mohamed Amine Zghal <[email protected]> Co-authored-by: Yinon Ehrlich <[email protected]> Co-authored-by: Matthew Brandman <[email protected]> Co-authored-by: claude[bot] <209825114+claude[bot]@users.noreply.github.com> Co-authored-by: Douwe Maan <[email protected]> Co-authored-by: Alex Enrique <[email protected]> Co-authored-by: Jerry Yan <[email protected]> Co-authored-by: Claude <[email protected]> Co-authored-by: Mayank <[email protected]> Co-authored-by: Alex Hall <[email protected]> Co-authored-by: Jerry Lin <[email protected]> Co-authored-by: Raymond Xu <[email protected]> Co-authored-by: kauabh <[email protected]> Co-authored-by: Victorien <[email protected]> Co-authored-by: Ethan Brooks <[email protected]> Co-authored-by: eballesteros <[email protected]>

ryx2 mentioned this pull request Aug 10, 2025

Support for gpt5's 'verbosity' option #2478

Closed

DouweM requested changes Aug 12, 2025

View reviewed changes

pydantic_ai_slim/pydantic_ai/models/openai.py Outdated Show resolved Hide resolved

pydantic_ai_slim/pydantic_ai/models/openai.py Outdated Show resolved Hide resolved

tests/models/test_openai_responses.py Outdated Show resolved Hide resolved

DouweM self-assigned this Aug 12, 2025

DouweM added the awaiting author revision label Aug 12, 2025

ryx2 and others added 5 commits August 12, 2025 14:48

ryx2 requested a review from DouweM August 13, 2025 00:35

Merge branch 'main' into feat/openai-verbosity-support

909cc16

DouweM requested changes Aug 13, 2025

View reviewed changes

pydantic_ai_slim/pydantic_ai/models/openai.py Outdated Show resolved Hide resolved

DouweM reviewed Aug 13, 2025

View reviewed changes

uv.lock Show resolved Hide resolved

ryx2 and others added 2 commits August 13, 2025 11:14

Update pydantic_ai_slim/pydantic_ai/models/openai.py

b09898b

good suggestion Co-authored-by: Douwe Maan <[email protected]>

ryx2 force-pushed the feat/openai-verbosity-support branch from 7c42321 to 099edda Compare August 13, 2025 18:36

ryx2 added 4 commits August 13, 2025 11:37

Merge branch 'main' into feat/openai-verbosity-support

da07a88

do the uv lock add only after make install

e71c25c

do the uv lock add only after make install and deleting it first

7ef085a

updated my uv from 0.5.3 to 0.8.9

37149a9

ryx2 requested a review from DouweM August 13, 2025 19:11

DouweM reviewed Aug 13, 2025

View reviewed changes

pydantic_ai_slim/pyproject.toml Outdated Show resolved Hide resolved

Merge branch 'main' into feat/openai-verbosity-support

5a4e09a

ryx2 and others added 2 commits August 13, 2025 13:20

Merge remote-tracking branch 'upstream/main' into feat/openai-verbosi…

ed2ce7f

…ty-support

Use upstream pyproject.toml from main

9267444

🤖 Generated with [Claude Code](https://claude.ai/code) Co-Authored-By: Claude <[email protected]>

ryx2 added 3 commits August 13, 2025 13:22

Merge branch 'feat/openai-verbosity-support' of https://github.com/ry…

a7af490

…x2/pydantic-ai into feat/openai-verbosity-support

revert pyproject change + newline lint issue

487245c

Delete pydantic_ai_slim/pyproject.toml.main

1ab457c

ryx2 requested a review from DouweM August 13, 2025 20:35

DouweM merged commit 5f99595 into pydantic:main Aug 13, 2025
64 of 66 checks passed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Add support for OpenAI verbosity parameter in Responses API #2493

Add support for OpenAI verbosity parameter in Responses API #2493

Uh oh!

ryx2 commented Aug 10, 2025

Uh oh!

medaminezghal commented Aug 11, 2025

Uh oh!

Uh oh!

Uh oh!

Uh oh!

ryx2 commented Aug 13, 2025

Uh oh!

ryx2 commented Aug 13, 2025

Uh oh!

Uh oh!

Uh oh!

Uh oh!

DouweM commented Aug 13, 2025

Uh oh!

ryx2 commented Aug 13, 2025

Uh oh!

ryx2 commented Aug 13, 2025

Uh oh!

Uh oh!

DouweM commented Aug 13, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Add support for OpenAI verbosity parameter in Responses API #2493

Add support for OpenAI verbosity parameter in Responses API #2493

Uh oh!

Conversation

ryx2 commented Aug 10, 2025

Summary

Test plan

Implementation Details

Uh oh!

medaminezghal commented Aug 11, 2025

Uh oh!

Uh oh!

Uh oh!

Uh oh!

ryx2 commented Aug 13, 2025

Uh oh!

ryx2 commented Aug 13, 2025

Uh oh!

Uh oh!

Uh oh!

Uh oh!

DouweM commented Aug 13, 2025

Uh oh!

ryx2 commented Aug 13, 2025

Uh oh!

ryx2 commented Aug 13, 2025

Uh oh!

Uh oh!

DouweM commented Aug 13, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants