Cohere Command Light (legacy)
Cohere's fast lightweight chat model (deprecated Sep 2025). Kept as comparison tombstone.
Cohere Command Light (legacy) is text & chat AI model from Cohere, priced at €0.300 per 1M input tokens with a 4.1K tokens context window.
0.7
Pricing
API Integration
Use our OpenAI-compatible API to integrate Cohere Command Light (legacy) into your application.
npm install railwailimport railwail from "railwail";
const rw = railwail("YOUR_API_KEY");
// Simple — just pass a string
const reply = await rw.run("command-light", "Hello! What can you do?");
console.log(reply);
// With message history
const reply2 = await rw.run("command-light", [
{ role: "system", content: "You are a helpful assistant." },
{ role: "user", content: "Explain quantum computing simply." },
]);
console.log(reply2);
// Full response with usage info
const res = await rw.chat("command-light", [
{ role: "user", content: "Hello!" },
], { temperature: 0.7, max_tokens: 500 });
console.log(res.choices[0].message.content);
console.log(res.usage);Deep dive — Cohere's Cohere Command Light (legacy)
Cohere is an enterprise-focused LLM lab founded in 2019 by Aidan Gomez (co-author of 'Attention Is All You Need'), Ivan Zhang and Nick Frosst (former Google Brain). Headquartered in Toronto with offices in San Francisco, London and New York, Cohere has positioned itself as an enterprise alternative to OpenAI and Anthropic with a focus on retrieval-augmented generation, tool use, multilingual quality and deployment flexibility (SaaS, AWS Bedrock, Azure, Oracle, on-prem via Cohere North). Major investors include Nvidia, Cisco, Salesforce, Oracle and Inovia Capital, with total funding above $1B and a 2024 valuation of approximately $5.5B. The Command family is Cohere's flagship LLM line: Command (2023), Command Light (2023, the small/fast tier), Command R and R+ (2024) and command-r7b (2024).
Visit Cohere →Command Light is the small, fast variant of Cohere's first-generation Command family, launched in 2023 and predating the Command R / R+ release. It is a closed-weight dense decoder-only transformer optimised for low latency and high throughput on simple text tasks — classification, extraction, short summarisation and routing. Cohere has not published exact architecture details (layers, heads, hidden size) or training data composition; the model is distributed only through the Cohere hosted API and was kept intentionally light for high-volume cost-sensitive workloads. Command Light was effectively deprecated when Command R launched in March 2024, though Cohere has maintained backwards-compatible pinned aliases (command-light-nightly) for legacy customers during their support window.
- Parameters
- Undisclosed (estimated ~6B based on contemporaneous tier siblings)
- Context
- 4.1K tokens
- Very low latency and high throughput for simple text tasks
- Cheap classification and extraction
- Short summarisation and routing decisions
- Stable production behaviour on pinned aliases
- Hosted-only via Cohere API
- Multilingual support for major business languages (limited vs. Command R)
- Effectively deprecated — Cohere directs new builds to Command R / Command R7B
- Best for: legacy pipelines on a planned migration path to Command R.
Not publicly disclosed. Cohere-curated web corpora plus proprietary instruction data is the standard 2023 Command-family recipe. Knowledge cutoff approximately early 2023.
License: Cohere Terms of Service. Closed weights — commercial use only via the hosted API. No self-hosting or fine-tuning of base weights for external customers (light custom-model training is available as a managed service).
Known limitations
- Officially deprecated tier — sunset on Cohere's published schedule
- No tool use or function calling (introduced in Command R)
- Weak on multi-step reasoning and long-form generation
- Multilingual coverage narrower than Command R / Aya
- Short 4K context window
- Closed weights — no self-hosting option
Frequently asked questions
Related Models
View all Text & ChatClaude Opus 4
Anthropic's most powerful model. Exceptional at complex analysis, agentic tasks, and extended reasoning.
Claude Sonnet 4
Anthropic's most capable model. Excellent for complex analysis, coding, math, and creative writing.
DeepSeek V3.1
DeepSeek's refreshed V3.1 release. 671B MoE / 37B active. Tops open-weights leaderboards on coding and reasoning.
DeepSeek V4 Pro
DeepSeek's April 2026 flagship. 1.6T MoE / 49B active params, 1M context, rivals top closed-source models on STEM and coding at a fraction of the price.
Start using Cohere Command Light (legacy) today
Get started with free credits. No credit card required. Access Cohere Command Light (legacy) and 100+ other models through a single API.