# Moonshot AI vs Cohere

> Kimi K3 is the most capable open-weight model for long coding runs. Cohere's Command models are smaller and aimed at private enterprise RAG.

Canonical: https://www.subconscious.dev/compare/moonshot-ai-vs-cohere · By The Subconscious Team · Updated September 30, 2026

## How they compare

Moonshot's Kimi K3 is a 2.8 trillion parameter mixture-of-experts model with a 1M context and native vision, and Vals AI scored it 93.4% on SWE-bench Verified, fourth overall. The hosted API costs $3 in and $15 out with cached input at $0.30. Cohere's Command A+ is a 218B model with 25B active and a 128K window, and Cohere says it trails the latest open models on agentic coding. Command A lists at $2.50 in and $10 out with 256K context. For repo-scale coding agents, Kimi K3 is the stronger model, though it always thinks and runs around 33 tokens per second.

Cohere wins on practicality. Self-hosting Kimi K3 takes a 64+ accelerator cluster, and its custom license adds a commercial agreement above $20M in hosting revenue. Command A+ runs on two H100s or one B200 in 4-bit form under Apache 2.0, and Cohere supports private VPC and on-prem installs with fine-tuning. Moonshot paused new API subscriptions after K3 demand overran its GPUs, while Cohere sells through Bedrock, Azure AI Foundry and OCI. Cohere also offers Embed 4 and Rerank 4. Choose Kimi K3 for hard coding, Cohere for manageable enterprise deployments.

## What each one does

### Moonshot AI

Moonshot AI is the Beijing lab behind the Kimi models. Its flagship Kimi K3 launched July 16, 2026 as a 2.8 trillion parameter mixture-of-experts model that activates 16 of 896 experts per token, with native vision and a 1M token context. It is the first open model in the 3T class, and full weights landed on Hugging Face on July 27. The hosted API costs $3 in and $15 out per million tokens, with cached input at $0.30, and it runs through an OpenAI-compatible endpoint, Kimi Code in the terminal, OpenRouter and Cloudflare Workers AI.

### Cohere

Cohere is a Toronto-based lab that sells models and platforms to banks, governments and large enterprises rather than consumers. Its generative line is the Command family. Command A+, released May 20, 2026, is a 218B-parameter mixture-of-experts model with 25B active, published under Apache 2.0 with a 128K context window, and it combines reasoning, vision, translation and tool use in one set of weights. Command A has a 256K window and lists at $2.50 in and $10 out per million tokens, while Command R7B costs $0.0375 in. June 2026 added North Mini Code, a 30B Apache 2.0 coding model, and the lineup also includes Aya multilingual models and Transcribe for speech.

## Which is best, and when

### Choose Moonshot AI for

- Long-horizon coding on huge repositories
- Near-frontier coding scores on open weights
- Document-heavy research with 1M context

### Choose Cohere for

- Self-hosting on two H100s
- Permissive Apache 2.0 licensing at any scale
- Enterprise RAG with reranking

## At a glance

| Attribute | Moonshot AI | Cohere |
|---|---|---|
| Model access | Open weights, custom license | Closed, plus open Command A+ |
| Flagship models | Kimi K3, Kimi K2.6 | Command A+, Command A, Embed 4, Rerank 4 |
| Speed | ~33 tok/s on Kimi K3 | 375 tok/s on Command A+ W4A4, per Cohere |
| Price | $3 in, $15 out (Kimi K3) | $0.0375–$2.50 in, $0.15–$10 out per 1M |
| Customization | Open weights to fine-tune | Enterprise fine-tuning, incl. private |
| Deployment | API, Kimi Code, OpenRouter | API, Bedrock, Azure, OCI, VPC, on-prem |
| Long context | 1M | 256K on Command A; 128K on A+ |

## FAQ

### What is the difference between Moonshot AI and Cohere?

Kimi K3 is the most capable open-weight model for long coding runs. Cohere's Command models are smaller and aimed at private enterprise RAG.

### When should I choose Moonshot AI over Cohere?

Long-horizon coding on huge repositories; Near-frontier coding scores on open weights; Document-heavy research with 1M context.

### When should I choose Cohere over Moonshot AI?

Self-hosting on two H100s; Permissive Apache 2.0 licensing at any scale; Enterprise RAG with reranking.

### Is Moonshot AI or Cohere cheaper?

Moonshot AI: $3 in, $15 out (Kimi K3). Cohere: $0.0375–$2.50 in, $0.15–$10 out per 1M. The cheaper choice depends on the model and workload.

### Which has more context, Moonshot AI or Cohere?

Moonshot AI: 1M. Cohere: 256K on Command A; 128K on A+.

## Running long-horizon agents?

If your agents run past 200K tokens, compare both against Subconscious: [Subconscious vs Moonshot AI](https://www.subconscious.dev/compare/subconscious-vs-moonshot-ai.md), [Subconscious vs Cohere](https://www.subconscious.dev/compare/subconscious-vs-cohere.md).

Full profiles: [Moonshot AI](https://www.subconscious.dev/providers/moonshot-ai.md), [Cohere](https://www.subconscious.dev/providers/cohere.md).