AI models released in May 2023
10 releases from 7 labs in May 2023, including CodeT5+, VisualGLM-6B and Claude and Claude Instant (100K context). Every entry is dated to the lab's official announcement and links to it.
- Releases
- 10
- Labs
- 7
- Open weights
- 5
May 2023
10 releasesCodeT5+
Family of open code language models with improved architectures and training, using a shallow encoder and deep decoder built on frozen code LLMs for efficient scaling.
- Modalities
- text → text
- Parameters
- 220M to 16B
- License
- BSD-3-Clause
- At launch
- Open weights
- Evaluated on more than 20 code benchmarks
- Instruction-tuned InstructCodeT5+ 16B also released
- Released in sizes from 220M to 16B
VisualGLM-6B
VisualGLM-6B is Z.ai's open multimodal dialogue model that supports images, Chinese, and English. It builds on the 6.2B-parameter ChatGLM-6B language model and bridges vision and language with a BLIP2-Qformer.
- Modalities
- text, image → text
- License
- VisualGLM-6B License
- At launch
- Open weights
- Pre-trained on 30M Chinese and 300M English image-text pairs
- Language model based on ChatGLM-6B
- Code under Apache 2.0; weights under a separate model license
Claude and Claude Instant (100K context)
Anthropic expanded Claude's context window from 9K to 100K tokens, about 75,000 words, letting businesses submit hundreds of pages of material or long transcripts in a single prompt. The 100K versions were made available through the API.
- Modalities
- text → text
- Context
- 100K context
- At launch
- API
- Context window expanded from 9K to 100K tokens
- Available in the Anthropic API
- Anthropic demonstrated Claude Instant reading The Great Gatsby (72K tokens) in a single prompt
InstructBLIP
Vision-language instruction-tuning framework built on BLIP-2 models, released with state-of-the-art zero-shot generalization across a wide range of vision-language tasks.
- Modalities
- text, image → text
- At launch
- Open weights
- Released in LAVIS with Vicuna 7B and 13B and FlanT5-XL and FlanT5-XXL variants
- Vicuna variants used frozen Vicuna v1.1 weights that users prepared separately
Chirp
Chirp is Google Cloud's speech-to-text foundation model, a version of the 2B-parameter Universal Speech Model supporting over 100 languages. It launched in preview on Vertex AI at Google I/O 2023.
- Modalities
- audio → text
- Parameters
- 2B
- At launch
- API, Cloud platforms
- Trained on millions of hours of audio
- Available in preview to everyone with a Google Cloud account at launch
- Generally available through the Speech-to-Text v2 API from August 2023
PaLM 2
PaLM 2 is Google's second-generation large language model, trained more heavily on multilingual text and source code, and offered in several sizes for uses ranging from mobile devices to the cloud.
- Modalities
- text → text
- At launch
- API, Cloud platforms, Consumer app
- Offered in four sizes: Gecko, Otter, Bison and Unicorn
- Pre-training spanned multilingual text in more than 100 languages
- Opened to developers through the PaLM API and to Google Cloud customers in Vertex AI
- Powered Bard and over 25 products and features announced at Google I/O 2023
MPT-7B
First MosaicML Foundation Series model, a decoder-only transformer trained from scratch on 1T tokens of text and code and licensed for commercial use, released by MosaicML before Databricks acquired the company in July 2023.
- Modalities
- text → text
- Parameters
- 6.7B
- License
- Apache-2.0
- At launch
- Open weights
- Retired
- Trained on the MosaicML platform in 9.5 days for about $200k
- StoryWriter-65k+ variant fine-tuned with a 65k-token context
- Instruct variant under CC BY-SA 3.0, Chat variant non-commercial
Databricks retired MPT-7B and MPT-7B Instruct from pay-per-token serving on August 30, 2024 and from provisioned throughput on December 19, 2025.
Source: Introducing MPT-7B: A New Standard for Open-Source, Commercially Usable LLMs | Databricks BlogRetirement: AI models maintenance policy | Databricks on AWS
Midjourney V5.1
Update to Midjourney's version 5 model with a stronger default aesthetic for simple prompts, high coherency, more accurate natural language interpretation and fewer unwanted artifacts and borders.
- Modalities
- text, image → image
- At launch
- Consumer app
- Supports repeating patterns with --tile
- Default model from May 2023 until June 22, 2023
- Generates 1024x1024 images
CodeGen2
Second generation of Salesforce's CodeGen program synthesis models, adding infill sampling and support for more programming languages than the original CodeGen family.
- Modalities
- text → text
- Parameters
- 1B, 3.7B, 7B and 16B
- License
- Apache-2.0
- At launch
- Open weights
- Released in 1B, 3.7B, 7B and 16B sizes
- Introduced special tokens for infill sampling
- Released under Apache 2.0
Cohere Rerank (v2.0)
Cohere's first Rerank endpoint, a model that scores the relevance of each search result to a query so it can be added as a final stage to keyword or semantic search.
- Modalities
- text → score
- At launch
- API
- Offered as rerank-english-v2.0 and rerank-multilingual-v2.0
- Supports more than 100 languages in the multilingual version
- Integrates with Elasticsearch, OpenSearch and Solr pipelines
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 May 2023." The AI Engineering, updated May 20, 2023. https://theaiengineering.com/releases/2023/05
HTML link
<a href="https://theaiengineering.com/releases/2023/05">AI models released in May 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