Cohere
39 models across 9 families · Latest: Cohere Transcribe Arabic (07-2026) (2026-07)
Empowering developers with advanced language AI.
Cohere's portfolio covers 36 active models across 9 current families, spanning coding, rag, and agents. Open a model detail page to compare provider routes and sourced benchmarks.
Covers 7 workload areas across 36 active tracked models; last verified 2026-07-13.
Use it for
- Teams evaluating coding, rag, and agents across this lab's releases
- Comparing model families before committing to a flagship
- Migration and pricing follow-ups across 36 tracked models
Do not use it for
- Choosing a hosting provider without opening a model page for price ladders
Active models
36
Current models from this lab, excluding deprecated ones
Active families
9
Current model families from this lab
Open catalog
16 open
3 open source / 13 open weights
Lowest output price
$1.50 /1M
Cheapest tracked output across active models, per 1M tokens
Latest dated release
2026-07-07
Cohere Transcribe Arabic (07-2026)
Freshness
2026-07-13
Researched 28d ago
Information
Release cadence
Showing 5 recent dated releases (full timeline below). Latest: Cohere Transcribe Arabic (07-2026) (2026-07-07).
Where this lab wins
- Coding: 2 tracked models with SWE-bench / HumanEval-style scores.
- RAG: 10 tracked models with ruler / needle retrieval benchmarks.
- Agentic: 7 tracked models with BFCL, tau-bench, and SWE-bench tool-use coverage.
- Long-context: 11 tracked models with context-token or InfiniteBench-class signal.
Flagship quality / price signal
Flagship: Command R+ (best sourced coding quality-per-dollar in this portfolio).
Quality-per-dollar unavailable for this flagship — benchmark coverage or output token pricing is still missing.
Cohere is a Canadian AI company founded in 2022. Empowering developers with advanced language AI. Cohere ships 9 model families totaling 39 models, with the most recent release Cohere Transcribe Arabic (07-2026) in 2026-07. Notable families include Cohere Transcribe, North, and Command. Use it as a stable reference for lab background, release coverage, and follow-up model pages as they are added. View official API endpoints, benchmark performance, and coding/agent fit for every Cohere model.
About
Cohere empowers developers with advanced language AI models and tools, making it easier to integrate natural language processing capabilities into a wide range of applications. Their pre-trained models and APIs can be leveraged by software engineers and SaaS executives to build innovative solutions in industries such as customer support, content creation, and sentiment analysis. Cohere For AI is their research arm focused on open science.
Featured models
| Model | Released | Context | Input price ($/1M) | Output price ($/1M) | License | Openness |
|---|---|---|---|---|---|---|
| Cohere Transcribe Arabic (07-2026) | 2026-07-07 | - | - | - | Apache 2.0 | Open source |
| North Mini Code 1.0 | 2026-06-09 | 256k | - | - | Apache 2.0 | Open source |
| Command A+ | 2026-05-20 | 128k | - | - | Proprietary | Proprietary |
Model families
Recent releases
- Cohere Transcribe Arabic (07-2026)- 2026-07-07
- North Mini Code 1.0- 2026-06-09
- Command A+- 2026-05-20
- Cohere Transcribe (03-2026)- 2026-03-01
- Command A Translate (08-2025)- 2025-08-01
Top comparisons
FAQ
Who founded Cohere and when?
Cohere was founded in 2022 and is associated with Toronto, Ontario, Canada.
What models has Cohere released?
Cohere ships 39 models across 9 families: Cohere Transcribe, North, and Command.
Is Cohere's technology open source?
Some Cohere models are open-weight (Cohere Transcribe Arabic (07-2026), Cohere Transcribe (03-2026), and North Mini Code 1.0); others are proprietary (Command A+, Command A, and Cohere Embed v4.0).
Where is Cohere headquartered?
Cohere is headquartered in Toronto, Ontario, Canada.
What is Cohere known for?
Empowering developers with advanced language AI. Its most prominent tracked family is Cohere Transcribe.
How can I access Cohere's models?
Cohere's models are available via Cohere API, AWS Bedrock, DeepInfra, IBM watsonx, and Microsoft Foundry.
Explore related pages
Last reviewed: 2026-07-13. Data sourced from public lab announcements and provider documentation.