AI models released in April 2023

7 releases from 6 labs in April 2023, including DeepFloyd IF, Eleven Multilingual v1 and Bark. Every entry is dated to the lab's official announcement and links to it.

Releases
7
Labs
6
Open weights
5

April 2023

7 releases
  1. Stability AIImageOpen weights

    DeepFloyd IF

    Cascaded pixel diffusion text-to-image model from Stability AI's DeepFloyd research lab, using a T5-XXL text encoder to render coherent text in images, released under a non-commercial research license.

    Modalities
    text → image
    License
    DeepFloyd IF License Agreement
    At launch
    Open weights
    • Research release on a non-commercial, research-permissible license
    • Uses T5-XXL-1.1 as the text encoder
    • Supports zero-shot image-to-image translation
    Source: Stability AI releases DeepFloyd IF, a powerful text-to-image model that can smartly integrate text into images
  2. ElevenLabsSpeechProprietary

    Eleven Multilingual v1

    Speech synthesis model building on Eleven Monolingual v1 that added seven languages: French, German, Hindi, Italian, Polish, Portuguese and Spanish.

    Modalities
    text → audio
    At launch
    Consumer app, API
    • Seven new languages
    • Available across all subscription plans
    • Selectable in the Speech Synthesis panel
    Source: Eleven Multilingual v1
  3. SunoSpeechOpen weights

    Bark

    Transformer-based text-to-audio model that generates realistic multilingual speech as well as music, background noise, simple sound effects and nonverbal sounds like laughing and sighing.

    Modalities
    text → audio
    License
    MIT
    At launch
    Open weights
    • Relicensed under MIT for commercial use on May 1, 2023
    • Pretrained checkpoints on GitHub and Hugging Face
    • Outputs limited to about 13 to 14 seconds per generation
    Source: suno-ai/bark: Text-Prompted Generative Audio Model
  4. Stability AILLMOpen weights

    StableLM (Alpha)

    First Stable LM release, 3B and 7B base language models trained on an experimental dataset built on The Pile with 1.5 trillion tokens, plus research-only instruction-tuned variants.

    Modalities
    text → text
    Context
    4K context
    Parameters
    3B and 7B
    License
    CC-BY-SA-4.0
    At launch
    Open weights
    • Base models usable commercially under CC BY-SA 4.0
    • Trained with a 4,096-token sequence length
    • Instruction-tuned research models released alongside
    Source: Stability AI Launches the First of its Stable LM Suite of Language Models
  5. DatabricksLLMOpen weights

    Dolly 2.0 (dolly-v2-12b)

    12B instruction-following model based on EleutherAI's Pythia family, fine-tuned only on databricks-dolly-15k, a human-written instruction dataset licensed for research and commercial use.

    Modalities
    text → text
    Parameters
    12B
    License
    MIT
    At launch
    Open weights
    • Dataset written by more than 5,000 Databricks employees in March and April 2023
    • Training code, dataset and weights released for commercial use
    • databricks-dolly-15k licensed under CC BY-SA 3.0
    Source: Dolly: Open Instruction-Tuned LLM | Databricks Blog
  6. MetaMultimodalOpen weights

    Segment Anything Model (SAM)

    Segment Anything Model is a promptable image segmentation foundation model that can segment objects from clicks or boxes and transfer zero-shot to new image domains.

    Modalities
    image → image
    License
    Apache-2.0
    At launch
    Open weights
    • Released with SA-1B, over 1.1 billion masks on about 11 million images
    • Produces a segment in about 50 milliseconds in a browser once the image embedding is computed
    Source: Introducing Segment Anything: Working toward the first foundation model for image segmentation
  7. MidjourneyImageProprietary

    Niji 5

    Anime and illustration model from the Niji series, built with Spellbrush and tuned for emotional, dramatic and artistic illustrations with cleaner light and shadow, precise small details and better coherency on subjects such as hands.

    Modalities
    text, image → image
    At launch
    Consumer app
    • Developed in collaboration with Spellbrush
    • Introduced --s to tone the Niji aesthetic up or down
    • Expressive, cute and scenic --style options followed in the weeks after launch
    Source: Introducing Niji 5!

Cite and reuse this data

These release records are free to reuse under CC BY 4.0, including commercially. Credit The AI Engineering and link to this page's URL.

Dates are the labs' official announcement dates, and every record links to its source. List fields in the CSV are separated by " | ".

Citation

The AI Engineering Editors. "AI models released in April 2023." The AI Engineering, updated Apr 28, 2023. https://theaiengineering.com/releases/2023/04

HTML link

<a href="https://theaiengineering.com/releases/2023/04">AI models released in April 2023</a> (The AI Engineering)
Full dataset
1,167 releases from 55 labs
Coverage
to
Updated
License
CC BY 4.0
Publisher
The AI Engineering

Archive