66 models with profiles. Classification describes capabilities, not a quality ranking.
Proprietary OpenAI model released on October 26, 2026 for interpreting images, using applications and writing text. It accepts text and images and returns text; generating images through a tool is not a native output modality. It is available through the API as gpt-6.1-terra and in ChatGPT Work and Codex depending on the plan. It is not released as a downloadable checkpoint.
Text and reasoning · Paid proprietary↗ TencentTencent Hunyuan checkpoint for simultaneous Chinese-English speech translation, included in the October 26, 2026 announcement. It receives incremental audio and produces translated text. The Interleave variant also retains source text; this is a different output, not a quantization. Repository creation on October 22 is not presented as the release date.
Voice and audio · Open weights↗ TencentTencent Hunyuan checkpoint for simultaneous Chinese-English speech translation, included in the October 26, 2026 announcement. It receives incremental audio and produces translated text. The Interleave variant also retains source text; this is a different output, not a quantization. Repository creation on October 22 is not presented as the release date.
Voice and audio · Open weights↗ inclusionAIinclusionAI vision model based on Qwen3.8-27b. It learns to locate interface elements and inspect image regions through cropping, transferring that strategy to general visual questions. The usable October 26 release includes twelve weight shards and a full model card; the empty repository earlier that morning was not considered a release. Its output consists of text and crop actions, not generated images.
Vision and documents · Open weights↗ BaiduBaidu model for extracting content from multi-page documents, announced on October 27, 2026. SR-SWA retains working references within a finite window to reduce attention cost growth. Input consists of page images in order and output is structured text. PDFs require prior conversion into images. Repository creation the previous day does not replace the announcement date.
Vision and documents · Open weights↗ NVIDIAResearch checkpoint published on October 20, 2026, for surgical robotics. A single model offers three modes: direct action-to-video dynamics, inverse video-to-action dynamics, and a policy conditioned on image, text, and history. It derives from Cosmos3-Nano, with a Qwen3-VL reasoning tower and fine-tuning of the generative tower. It is not three independent models nor an authorized medical device.
Robotics · Open weights↗ NVIDIADistilled causal variant of Cosmos3-Nano-Transfer-Auto, published by NVIDIA on October 20, 2026, for closed-loop driving simulation. It produces synchronized multiview video and LiDAR from controls and text. It supports up to eleven cameras. The ability to respond to controls during a sequence distinguishes it from the bidirectional generator, but it does not constitute an autonomous driving policy.
Robotics · Open weights↗ NVIDIANVIDIA causal model for visually simulating bimanual manipulation and camera navigation, published on October 20, 2026. It receives an initial image, robot actions, camera trajectory, and optional text. Its output is video, not actions: it should not be confused with a control policy. The checkpoint declares an AgiBot body; other morphologies require specific validation.
Robotics · Open weights↗ NVIDIANVIDIA distilled causal model published on October 20, 2026, to convert visual simulation controls into RGB video. Geometry, physics, and ground truth still belong to the classic simulator; the network provides visual appearance. It accepts text and depth, segmentation, edge, or blur controls. It is not a policy that produces robot actions nor a certified physics engine.
Robotics · Open weights↗ NVIDIANVIDIA bidirectional model for generating driving scenes with synchronized multiview video and LiDAR. Published on October 20, 2026, it combines text, WSM world controls, and an optional initial image. It supports up to eleven standard or fisheye cameras and up to 300 frames per view. It is batch generation, not the causal simulation model Sim-Auto.
Robotics · Open weights↗ Mistral4B parameter offline transcription variant announced by Mistral on October 20, 2026. It converts audio to text and supports diarization and contextual vocabulary. The current specification includes 50 languages. It is a distinct model from Small and Mini Realtime, with its own price and hosted access in public preview. The declared size does not imply that its weights are available for download.
Voice and audio · Paid proprietary↗ MistralStreaming audio transcriber presented by Mistral on October 20, 2026. The current guide lists 45 languages and dialects. It uses WebSocket, delivers incremental text, and adds turn-ending detection, speaker tags, and context_bias. It does not generate voice nor automatically share Small's offline capabilities. The versioned identifier allows distinguishing it from aliases that may change versions.
Voice and audio · Paid proprietary↗ MistralVoice synthesis model presented on October 20, 2026. The current specific guide declares 35 languages and more than 50 dialects; the old count of nine languages from the model card is not used as the current limit. It converts text to audio, with an optional voice reference. Batch and streaming responses use the same model; they are not duplicated as distinct checkpoints.
Voice and audio · Paid proprietary↗ MistralMistral transcriber presented on October 20, 2026, in public preview. It accepts audio files and returns text, with diarization and contextual bias for specialized vocabulary. The language guide declares 50 languages for offline transcription. Small and Mini are different variants with their own rates; this profile uses the versioned identifier voxtral-small-transcribe-3-0, not a moving alias.
Voice and audio · Paid proprietary↗ NVIDIANVIDIA image-editing model, version 1.0.0 released on October 20, 2026. It transforms synthetic PCB component crops rendered in Omniverse into an optical-inspection lighting style. It derives from Qwen-Image-Edit with a fused LoRA fine-tune: the package includes the complete pipeline, not a separate adapter. Users supply the image; the English instruction is fixed, not a free-form prompt.
Image · Open weights↗ NVIDIANVIDIA embedding model with 1.55B parameters, released on October 20, 2026. It vectorizes text documents, page images or both, with text-only queries. Distilled from Embed 8B, it freezes the Nemotron 3 Embed 1B text tower and adds visual retrieval. Its vector has 2,048 dimensions; it does not automatically share the 8B model’s space.
Embeddings · Open weights↗ NVIDIANVIDIA embedding model with 8.38B parameters, released on October 20, 2026. It vectorizes text, page images or both for retrieval with text queries. It freezes the Nemotron 3 Embed 8B tower and adds vision through Pixtral. It retains that predecessor’s text embedding space; it produces 4,096-dimensional vectors, not conversational responses.
Embeddings · Open weights↗ NVIDIANVIDIA reranker with 8.38B parameters, released on October 20, 2026. It scores each text-query and document pair; documents may contain text, a page image or both. It complements initial retrieval by reranking candidates; it does not generate responses or embeddings. It combines a Pixtral vision encoder with a bidirectional Ministral3 tower.
Reranking · Open weights↗
A workflow for recording how an image, video, or audio file is created, modified, and published—and for interpreting provenance credentials with care.
29 Sep 2026 ↗
A protocol for comparing Stable Audio 3.0 with licensed-library sound effects—without mistaking an impressive sample for a reliable workflow. Includes listening criteria, time, editing, and rights.
26 Sep 2026 ↗
A generated shot can be convincing on its own and still be unusable when edited with others. This guide proposes a protocol for evaluating visual, spatial, and narrative continuity in generative-video sequences, recording their traceability, and deciding which clips are approved, which require intervention, and which must be discarded.
23 Sep 2026 ↗
A generative edit that is useful in production is not validated merely because the requested change looks correct. This guide proposes treating every retouch as a preservation contract: define what may change, what must remain, how to measure it, and when to require human review.
23 Sep 2026 ↗
Using a synthetic voice responsibly requires more than a checkbox or a convincing demo. This guide proposes an operating system for linking each audio asset to a specific authorization, controlling the risk of resemblance to identifiable people, disclosing its nature where appropriate, and removing assets when the terms of use change.
22 Sep 2026 ↗
C2PA credentials can provide cryptographically verifiable evidence about provenance statements and transformations of a file. On their own, they do not prove that an image is true, may be legally reused, or depicts people who gave their consent. This guide sets out a method for assessing, preserving, and communicating that evidence without overstating its conclusions.
22 Sep 2026 ↗