{ "@context": "https://schema.org", "@id": "https://aggr.aymericbeaumet.com/items/simonwillison-net/2026-08-22-llm-0-33/#webpage", "@type": [ "WebPage", "ArchiveComponent" ], "articleBody": "My highlights from this release:\n\n> - Upgraded to the OpenAI Python library 3.x and switched the HTTP client dependency from `httpx` to `httpx2`. [#1608](https://github.com/simonw/llm/issues/1608), [#1631](https://github.com/simonw/llm/pull/1631)\n\nI shipped a quick [0.32.1 fix](https://simonwillison.net/2026/Aug/21/llm/) for this yesterday, but this is the more comprehensive fix.\n\n> - `llm embed` and `llm embed-multi` now accept `--key`. The Python `EmbeddingModel.embed()`, `EmbeddingModel.embed_multi()`, `Collection.embed()` and `Collection.embed_multi()` methods accept `key=` too, passing the resolved per-call key to embedding plugins without changing shared model state. Existing plugins that read `self.key` continue to work through a compatibility fallback. Thanks, [ChrisJr404](https://github.com/ChrisJr404). [#757](https://github.com/simonw/llm/issues/757), [#1620](https://github.com/simonw/llm/pull/1620)\n\nThe embedding models now use the same pattern for keys that regular LLM models do.\n\n> - `llm prompt -t/--template` can now be repeated to combine templates in order. This allows model configuration and options from one template to be used with a prompt from another.\n\nThis unlocks a neat pattern where you can create templates that package a model with a set of default options:\n\n```\nllm -m gpt-5.6-luna -o reasoning_effort high --save lhigh\nllm \"Generate an SVG of a pelican riding a bicycle\" --save pelican\n# Combine and run the templates\nllm -t lhigh -t pelican\n```\n\n> - Reasoning-capable Responses API models now support a `reasoning_summary` option with `auto`, `concise`, and `detailed` values. This can be used with [llm openai endpoint --responses](https://llm.datasette.io/en/stable/other-models.html#openai-endpoint). [#1600](https://github.com/simonw/llm/issues/1600)\n\nThis is particularly useful for exercising different models that provide their own imitation of the OpenAI Responses API.\n", "authors": [], "captured_at": "2026-09-04T12:14:41.802665207+00:00", "category": "programming", "content_html": "

My highlights from this release:

\n
\n\n
\n

I shipped a quick 0.32.1 fix for this yesterday, but this is the more comprehensive fix.

\n
\n\n
\n

The embedding models now use the same pattern for keys that regular LLM models do.

\n
\n\n
\n

This unlocks a neat pattern where you can create templates that package a model with a set of default options:

\n
llm -m gpt-5.6-luna -o reasoning_effort high --save lhigh\nllm "Generate an SVG of a pelican riding a bicycle" --save pelican\n# Combine and run the templates\nllm -t lhigh -t pelican\n
\n
\n\n
\n

This is particularly useful for exercising different models that provide their own imitation of the OpenAI Responses API.

\n", "content_markdown": "My highlights from this release:\n\n> - Upgraded to the OpenAI Python library 3.x and switched the HTTP client dependency from `httpx` to `httpx2`. [#1608](https://github.com/simonw/llm/issues/1608), [#1631](https://github.com/simonw/llm/pull/1631)\n\nI shipped a quick [0.32.1 fix](https://simonwillison.net/2026/Aug/21/llm/) for this yesterday, but this is the more comprehensive fix.\n\n> - `llm embed` and `llm embed-multi` now accept `--key`. The Python `EmbeddingModel.embed()`, `EmbeddingModel.embed_multi()`, `Collection.embed()` and `Collection.embed_multi()` methods accept `key=` too, passing the resolved per-call key to embedding plugins without changing shared model state. Existing plugins that read `self.key` continue to work through a compatibility fallback. Thanks, [ChrisJr404](https://github.com/ChrisJr404). [#757](https://github.com/simonw/llm/issues/757), [#1620](https://github.com/simonw/llm/pull/1620)\n\nThe embedding models now use the same pattern for keys that regular LLM models do.\n\n> - `llm prompt -t/--template` can now be repeated to combine templates in order. This allows model configuration and options from one template to be used with a prompt from another.\n\nThis unlocks a neat pattern where you can create templates that package a model with a set of default options:\n\n```\nllm -m gpt-5.6-luna -o reasoning_effort high --save lhigh\nllm \"Generate an SVG of a pelican riding a bicycle\" --save pelican\n# Combine and run the templates\nllm -t lhigh -t pelican\n```\n\n> - Reasoning-capable Responses API models now support a `reasoning_summary` option with `auto`, `concise`, and `detailed` values. This can be used with [llm openai endpoint --responses](https://llm.datasette.io/en/stable/other-models.html#openai-endpoint). [#1600](https://github.com/simonw/llm/issues/1600)\n\nThis is particularly useful for exercising different models that provide their own imitation of the OpenAI Responses API.\n", "dateCreated": "2026-09-04T12:14:41.802665207+00:00", "date_created": "2026-08-22T17:01:16+00:00", "date_modified": "2026-08-22T17:01:16+00:00", "date_published": "2026-08-22T17:01:16+00:00", "description": "Release: llm 0.33 My highlights from this release: Upgraded to the OpenAI Python library 3.x and switched the HTTP client dependency from httpx to httpx2. #1608, #1631 I shipped a quick 0.32.1 fix for this yesterday, but this is the more…", "external_url": "https://simonwillison.net/2026/Aug/22/llm/", "git": { "commit": "ebb0941c5c06a2ab2377342d607535262a3e969e", "path": "items/simonwillison-net/2026/08/2026-08-22-llm-0-33.md", "permalink": "https://github.com/aymericbeaumet/aggr-instance/blob/ebb0941c5c06a2ab2377342d607535262a3e969e/items/simonwillison-net/2026/08/2026-08-22-llm-0-33.md" }, "id": "urn:aggr:item:4f6f6ad6ec5f1880d7849358da8681611176ca9b", "isBasedOn": { "@id": "https://simonwillison.net/2026/Aug/22/llm/" }, "mainEntity": { "@id": "https://simonwillison.net/2026/Aug/22/llm/", "@type": "CreativeWork", "archivedAt": { "@id": "https://aggr.aymericbeaumet.com/items/simonwillison-net/2026-08-22-llm-0-33/#webpage" }, "dateModified": "2026-08-22T17:01:16+00:00", "datePublished": "2026-08-22T17:01:16+00:00", "headline": "llm 0.33", "timeRequired": "PT2M", "url": "https://simonwillison.net/2026/Aug/22/llm/", "wordCount": 232 }, "name": "Archived snapshot: llm 0.33", "preview": null, "reading_minutes": 2, "replicated_at": null, "source": "simonwillison-net", "source_name": "Simon Willison's Weblog", "summary": "Release: llm 0.33 My highlights from this release: Upgraded to the OpenAI Python library 3.x and switched the HTTP client dependency from httpx to httpx2. #1608, #1631 I shipped a quick 0.32.1 fix for this yesterday, but this is the more comprehensive fix. llm embed and llm embed-multi now accept --key. The Python EmbeddingModel.embed(), EmbeddingModel.embed_multi(), Collection.embed() and Collection.embed_multi() methods accept key= too, passing the resolved per-call key to embedding plugins without changing shared model state. Existing plugins that read self.key continue to work through a compatibility fallback. Thanks, ChrisJr404. #757, #1620 The embedding models now use the same pattern for keys that regular LLM models do. llm prompt -t/--template can now be repeated to combine templates in order. This allows model configuration and options from one template to be used with a prompt from another. This unlocks a neat pattern where you can create templates that package a model with a set of default options: llm -m gpt-5.6-luna -o reasoning_effort high --save lhigh llm \"Generate an SVG of a pelican riding a bicycle\" --save pelican # Combine and run the templates llm -t lhigh -t pelican Reasoning-capable Responses API models now support a reasoning_summary option with auto, concise, and detailed values. This can be used with llm openai endpoint --responses. #1600 This is particularly useful for exercising different models that provide their own imitation of the OpenAI Responses API. Tags: annotated-release-notes, llm", "tags": [ "annotated-release-notes", "llm" ], "temporalCoverage": "2026-09-04T12:14:41.802665207+00:00", "title": "llm 0.33", "url": "https://aggr.aymericbeaumet.com/items/simonwillison-net/2026-08-22-llm-0-33/", "word_count": 232 }