Claude Haiku 4.5
TL;DR
PremiumFast completions, small edits, and quick lookups.
Available on
Not on the Go plan ($1/mo, open-source models only).
Switch with
Pick Claude Haiku 4.5 from the selector.
Intelligence Index
Haiku 4.5 vs the Claude 4 family
Opus 4.7
Sonnet 4.6
Haiku 4.5
Speed
~97
tokens / sec
Input
$1
per M tokens
Output
$5
per M tokens
Claude Haiku 4.5 in Command Code
Claude Haiku 4.5 is the fast, compact model in Anthropic's Claude 4 family — the cheapest pick in the Claude lineup. In Command Code it is the model you switch to when the task is small and the answer is needed fast.
Haiku 4.5 vs the Command Code lineup
Quality, speed, and pricing for Haiku 4.5 alongside the fastest peers.
| Model | Intelligence | Speed | Input $/M | Output $/M |
|---|---|---|---|---|
| Claude Haiku 4.5 | 37 | ~97 tok/s | $1.00 | $5.00 |
| GPT-5.4 Mini | 49 | ~164 tok/s | $0.75 | $4.50 |
| DeepSeek V4 Flash | 47 | ~82 tok/s | $0.14 | $0.28 |
| MiniMax M2.5 | Not yet scored | — | $0.27 | $0.95 |
| Claude Sonnet 4.6 | 52 | ~62 tok/s | $3.00 | $15.00 |
| Kimi K2.5 | 37 | ~35 tok/s | $0.60 | $3.00 |
What Haiku 4.5 is best for
Quick lookups, small edits, mechanical refactors, lint sweeps, formatting fixes, and short test generation. Anywhere short turnaround is more valuable than the strongest reasoning.
When to switch away from Haiku 4.5
Switch to Claude Sonnet 4.6
For multi-file refactors or anything that needs deep planning. Sonnet is the recommended default for general agent work.
Switch to Claude Opus 4.7
For the hardest reasoning work — large refactors, ambiguous specs, complex agent runs.
Switch to GPT-5.4 Mini
Comparable speed/cost tier with higher Intelligence Index (49 vs 37) at a slightly lower input price ($0.75 vs $1.00).
Switch to DeepSeek V4 Flash
For open-source economics at high volume. Flash costs $0.14 input vs $1.00 for Haiku, with comparable speed.
In Command Code: caching and taste-1
Two things change the experience of using this model inside Command Code versus calling it directly through the upstream API.
First, prompt caching is on by default. In an agent loop the same context is read across many steps; cache reads are billed at $0.10 per million tokens versus $1.00 for fresh input.
Second, taste-1 sits between the model and the agent loop, rewriting and reranking candidate edits to match your codebase conventions. Each plan ships with a taste-1 usage allowance that scales by tier (Go $100 → Ultra $10,000).
Plan availability
Premium model. Available on Pro ($15/mo), Max ($100/mo), Ultra ($200/mo), Teams ($40/mo per seat), and Enterprise. Not on the Go plan.
All Command Code models, ranked by quality and speed
Quality is the Intelligence Index — an aggregate score across reasoning, math, coding, and knowledge evaluations. Speed is reported output tokens per second. Models without a published score are noted.
| Model | Tier | Intelligence Index | Output speed |
|---|---|---|---|
| GPT-5.5 | Premium | 60 | ~65 tok/s |
| Claude Opus 4.7 | Premium | 57 | ~49 tok/s |
| GPT-5.4 | Premium | 57 | ~84 tok/s |
| GPT-5.3 Codex | Premium | 54 | ~72 tok/s |
| Kimi K2.6 | Open-source | 54 | ~40 tok/s |
| Claude Sonnet 4.6 | Premium | 52 | ~62 tok/s |
| DeepSeek V4 Pro | Open-source | 52 | ~35 tok/s |
| GLM-5 | Open-source | 50 | ~61 tok/s |
| GPT-5.4 Mini | Premium | 49 | ~164 tok/s |
| DeepSeek V4 Flash | Open-source | 47 | ~82 tok/s |
| Claude Haiku 4.5 | Premium | 37 | ~97 tok/s |
| Kimi K2.5 | Open-source | 37 | ~35 tok/s |
| Claude Opus 4.6 | Premium | Not yet scored | — |
| MiniMax M2.5 | Open-source | Not yet scored | — |
Switching models with /model
In an interactive Command Code session, run /model to open the model selector. Pick the model you want and it applies to this session and to future sessions until you change it again. Premium models require Pro or higher; open-source models are available on every plan, including Go.
cmd # start an interactive session
/model # open the selector and pick a modelPlans and pricing
Command Code is a subscription with model usage at API rates. Each plan ships with monthly LLM credits and a separate taste-1 usage allowance that scales by tier. Credits roll over and never expire. Auto top-up keeps you running if you go over.
| Plan | Price/mo | LLM credits | taste-1 usage | Models |
|---|---|---|---|---|
| Go | $1 | $10 | $100 | Open-source only |
| Pro | $15 | $30 | $500 | Open-source + premium |
| Max | $100 | $150 | $5,000 | Open-source + premium |
| Ultra | $200 | $300 | $10,000 | Open-source + premium |
| Teams | $40 / seat | Pooled | $1,000 | Open-source + premium |
| Enterprise | Custom | Custom | Custom | Custom pool, SSO, audit logs |
Frequently asked questions
Should I default to Haiku 4.5?
Only if your work is mostly quick lookups and small edits. Sonnet 4.6 is the recommended default for general agent runs; Haiku is the model to switch to with /model when latency wins.
Haiku 4.5 or GPT-5.4 Mini?
Mini scores higher on the Intelligence Index (49 vs 37) and runs faster (~164 vs ~97 tok/s). Pick Mini if you want OpenAI ergonomics; pick Haiku for the Claude family and its caching model.
Which Command Code model should I use?
Claude Sonnet 4.6 is the recommended default. Switch to GPT-5.5 (Intelligence Index 60) for the absolute hardest reasoning, or Claude Opus 4.7 / GPT-5.4 (both 57) for top-tier work at lower cost. For fast lookups, Claude Haiku 4.5 or GPT-5.4 Mini. For open-source, Kimi K2.6 leads the open-weights tier (Intelligence Index 54).
Can I mix Claude Haiku 4.5 with other models in a workflow?
Yes. Switch per session using /model. Common pattern: keep Sonnet 4.6 as the default and switch up to Opus 4.7 or down to Haiku 4.5 as the task calls for it.
Are open-source model prices fixed?
Open-source models are routed across multiple upstream providers for high availability. The price listed for each is the mean per-provider rate. Actual cost on a given request may vary slightly. The Usage page reflects the price charged.
Is Command Code free to try?
The Go plan starts at $1/mo with $10 in LLM credits and $100 of taste-1 usage. It covers open-source models only. Pro at $15/mo unlocks premium models with $30 in LLM credits and $500 of taste-1 usage.
Does Command Code train on my code?
No. Command Code does not train on your code or store your code snippets. taste-1 data is stored locally in your project directory.
Where can I track my usage?
The Usage page in Studio shows per-request cost, token counts, and which model ran. Settings > Billing lets you change plans, buy credits, or enable auto top-up.
Does Command Code replace my editor?
No. Command Code is editor-agnostic — it runs as a CLI and works alongside any editor (Cursor, VS Code, Zed, JetBrains, Neovim, etc.).
Ship code that matches your taste
Command Code is the AI coding agent that continuously learns your taste. Start for $1.