Slug: cohere · Theme: ai-infra · Date: 2026-09-21 (CT)
Sources preference: primary (cohere.com pricing, about, product, docs.cohere.com) over press.
| Model | Performance tier (examples shown) | Hourly / Monthly (where listed) |
|---|---|---|
| Embed 4 | Small | $4.00 / hr · $2,500 / mo |
| Embed 4 | Medium | $5.00 / hr · $3,250 / mo |
| Rerank 4 Pro | (tier row on page) | Monthly example $4,300 (page table; other tiers present) |
command-a-plus-05-2026), Command A (command-a-03-2025), Command R7B, Command A Translate / Reasoning / Vision, Command R / R+ (08-2024 live; older R/R+ aliases deprecated Sept 15, 2025). Chat endpoint; RAG / tool use / agents.embed-v4.0 (text + images + mixed/PDF; dims 256–1536) plus v3 English/multilingual light variants.rerank-v4.0-pro, rerank-v4.0-fast, plus v3.5 / v3 English & multilingual.parse-v5.0), Transcribe (incl. Arabic variant), Aya multilingual research models, North purpose-built models (e.g. North Small Translate, North Mini Code) on Chat; North Mini Code also via Model Vault.# Cohere — research brief **Slug:** cohere · **Theme:** ai-infra · **Date:** 2026-09-21 (CT) **Sources preference:** primary (cohere.com pricing, about, product, docs.cohere.com) over press. ## FACT (primary) ### Company - Cohere — enterprise AI: foundational models (Command, Embed, Rerank, Parse, Transcribe) plus **North** agentic workplace platform and **Model Vault** dedicated inference (cohere.com, about, docs). - Founded **2019 in Toronto** by Aidan Gomez (CEO), Nick Frosst, Ivan Zhang (cohere.com/about). - About page: between 2021–2024, four major rounds (Series A–D), raising **nearly $1 billion** (company claim). - **Cohere Labs** open-science initiative (2022); company claims **4,500+** community members and **100+** research papers (about). - **North** launched **2025** as turnkey agentic AI platform for workplace productivity (about / north pages). - Offices listed: Toronto, New York, London, San Francisco, Montreal, Paris, Seoul (about). ### Access & keys (cohere.com/pricing FAQ + docs rate-limits) - Signup creates a **Trial API key** — free, rate-limited; not for production/commercial use. - **Production API key** — pay-as-you-go; billed monthly or when outstanding balance hits **$250**. - Trial vs Production documented on pricing FAQ and docs.cohere.com/docs/rate-limits. - Personal accounts start with Trial only; Organizations share a billing portal (pricing FAQ). ### Model Vault (dedicated managed deployment — pricing page) | Model | Performance tier (examples shown) | Hourly / Monthly (where listed) | |-------|-----------------------------------|----------------------------------| | Embed 4 | Small | **$4.00 / hr** · **$2,500 / mo** | | Embed 4 | Medium | **$5.00 / hr** · **$3,250 / mo** | | Rerank 4 Pro | (tier row on page) | Monthly example **$4,300** (page table; other tiers present) | - Fully managed, isolated (no shared multi-tenant overhead); Fixed or Flex plans; integrates with North; self-serve via dashboard or contact sales (pricing). ### Products / model families (docs.cohere.com/docs/models, retrieved 2026-09-21) - **Command family:** Command A+ (`command-a-plus-05-2026`), Command A (`command-a-03-2025`), Command R7B, Command A Translate / Reasoning / Vision, Command R / R+ (08-2024 live; older R/R+ aliases deprecated Sept 15, 2025). Chat endpoint; RAG / tool use / agents. - **Embed:** `embed-v4.0` (text + images + mixed/PDF; dims 256–1536) plus v3 English/multilingual light variants. - **Rerank:** `rerank-v4.0-pro`, `rerank-v4.0-fast`, plus v3.5 / v3 English & multilingual. - **Parse** (`parse-v5.0`), **Transcribe** (incl. Arabic variant), **Aya** multilingual research models, **North** purpose-built models (e.g. North Small Translate, North Mini Code) on Chat; North Mini Code also via Model Vault. - Platforms: Cohere proprietary API; **Amazon SageMaker**, **Amazon Bedrock**, **Microsoft Azure AI Foundry**, **Oracle GenAI Service** (docs models page). ### North platform (cohere.com/north) - Agentic enterprise workspace: Discover / Create / Automate; connectors; private deploy in customer VPC, on-prem, or Model Vault. - Sales-led packaging (“Request a demo”); banking vertical callout (e.g. North for Banking / RBC quote on page). ### Legacy / FAQ rate card excerpts (cohere.com/pricing FAQ — FACT for listed SKUs) - Existing-customer **legacy** Command: $1.00 / $2.00 per 1M in/out; Command-light $0.30 / $0.60; Command R 03-2024 $0.50 / $1.50; Command R+ 04-2024 $3.00 / $15.00; Command R+ 08-2024 **$2.50 / $10.00**. - **Aya Expanse** (8B/32B) API: **$0.50 / $1.50** per 1M in/out. - Rerank billing unit: one **search** = one query with up to **100** documents; docs >500 tokens (incl. query) auto-chunked and each chunk counts as a document. ### Security posture (company marketing / about) - Enterprise security and data-protection framing; private deployment and Model Vault isolation called out on pricing and North pages. Specific cert names should be verified on security pages before citing as FACT checklist items. ## SECONDARY (press — do not treat as audited financials) - Jul 2024 Series D **~$500M** at **~$5.5B** valuation (TechCrunch / VentureBeat / Bloomberg-cited); investors reported include PSP, Cisco, AMD Ventures, Fujitsu, Nvidia, Salesforce Ventures, Oracle, EDC, Magnetar. - Aug 2025 additional **~$500M** at **~$6.8B** valuation led by Radical Ventures / Inovia (BetaKit) — SECONDARY. - Press ARR claims (e.g. Reuters ~$100M ARR) — **not used** as canvas metrics; unaudited. ## INFERENCE (labeled in canvas where used) - Teaching arc: **Trial / playground land → usage-based Command + Embed + Rerank Production API (developer PLG) → Model Vault / private deploy + North enterprise → agents/RAG stack + hyperscaler marketplaces**. - Cost structure mirrors frontier LLM vendors: GPU/cloud training & inference COGS, free Trial demand-gen, research (Cohere Labs + Command family), enterprise sales for North/private deploy, marketplace partner overhead. ## Sources - https://cohere.com/ - https://cohere.com/pricing - https://cohere.com/about - https://cohere.com/north - https://cohere.com/command - https://docs.cohere.com/docs/models - https://docs.cohere.com/docs/rate-limits - https://techcrunch.com/2024/07/22/cohere-raises-500m-to-beat-back-generative-ai-rivals/