Best Model for Claude Code, and Which Models It Can Run

Claude Code starts on Claude Opus 5.5 on Pro, Max, Team, Enterprise and API accounts, but Claude Sonnet 5.5 ($2/$10 per million tokens) is the best everyday model to run in it: second only to Opus 5.5 on Terminal-Bench 4.0, at half its price. Switch to Opus 5.5 ($4/$20) for the hardest tasks, and give subagents Claude Haiku 5.5 ($0.10/$0.50 up to 100K-token prompts). Non-Claude models also work through an Anthropic-compatible endpoint, and they cut the cost of routine steps further.

Read the full Claude Code review → · updated 2026-10-17

Asking which Claude model is best overall or for coding, outside Claude Code too? See the best Claude model ranking.

Which models can Claude Code use?

Pick a model with /model in a session, claude --model at startup, or the ANTHROPIC_MODEL environment variable. On the Anthropic API the aliases below resolve to the newest Claude models; on Bedrock, Google Cloud and Microsoft Foundry some resolve to older versions.

Alias	Model on the Anthropic API	Use it for
default	Opus 5.5 (Pro, Max, Team, Enterprise, API)	Clears any override and returns to your account's default
sonnet	Sonnet 5.5	Everyday coding at half Opus's price
opus	Opus 5.5	Hard refactors, tricky bugs, long autonomous runs
haiku	Haiku 5.5	Subagents and quick steps
opusplan	Opus 5.5 to plan, Sonnet 5.5 to execute	Planning quality with cheaper implementation
fable	Fable 5.1	Anthropic's premium tier for the longest tasks

CLAUDE_CODE_SUBAGENT_MODEL sets the model for subagents that are not assigned one, so a session can run on Sonnet or Opus while subagents run on Haiku.
Fable is never the account default; choosing it with /model saves it for later sessions.

Sources: Claude Code model configuration

#1Claude Sonnet 5.5Anthropic · $4.40/1M · 1M ctx

The best everyday pick: 61.8% on Terminal-Bench 4.0 in Claude Code, close to Opus 5.5, at half the price per token.

Overall 96/100 · Value 59Full breakdown →

#2Claude Opus 5.5Anthropic · $8.80/1M · 1M ctx

The quality ceiling for hard refactors, tricky bugs and long autonomous runs; #1 on Terminal-Bench 4.0 and Artificial Analysis.

Overall 98/100 · Value 51Full breakdown →

#3Claude Haiku 5.5Anthropic · $0.22/1M · 1M ctx

For subagents and the many small calls a session makes. Keep its prompts under 100K tokens, where it costs a fifth of its higher band.

Overall 82/100 · Value 97Full breakdown →

#4Claude Fable 5.1Anthropic · $22/1M · 1M ctx

Anthropic's premium tier. Worth testing only if your own tasks show a gain over Opus 5.5, which matches it on most work for less.

Overall 93/100 · Value 41Full breakdown →

Use cheaper non-Claude models in Claude Code

Claude Code accepts any endpoint that speaks Anthropic's Messages API: set ANTHROPIC_BASE_URL, a key and ANTHROPIC_MODEL. On our measured coding-agent traffic, 100 requests cost about $0.67 on DeepSeek V4.1 Flash and $1.29 on MiniMax M3, against $5.33 on Sonnet 5.5 and $10.66 on Opus 5.5 at list prices. The cheaper models are weaker on hard multi-step work, so the common setup keeps Claude for the hard steps and sends routine ones to a low-cost model. Standard Compute does that per request on one key: its router picks the cheapest model that can handle each step and saves frontier models such as Claude Fable 5.1 and GPT-6 Astra for the steps that need them, for a fixed monthly price from $19.

Run other models in Claude Code →DeepSeek in Claude Code →GLM 5.3 in Claude Code →Kimi K3 in Claude Code →MiniMax M3 in Claude Code →

What 100 Claude Code requests cost on each pick

A typical request sends 99,534 input tokens, 87% of them read from cache, and a median session runs 16 requests. Measured on pooled coding-agent requests through our gateway (79,301 requests from 75 users; Claude Code alone has fewer than 50 users, so its traffic is pooled), Sep 13 to Oct 12, 2026, priced at each vendor's list rates.

Model	100 requests	Median session	Month of regular use
Claude Sonnet 5.5	$5.33	$0.74	$77.42
Claude Opus 5.5	$10.66	$1.47	$155
Claude Haiku 5.5	$1.31	$0.096	$17.73
Claude Fable 5.1	$23.76	$3.26	$332

Month of regular use = the median user active on 5 or more days (1,478 requests in 30 days). Methodology · All models ranked by cost per request

Best model for Claude Code under $19, $49 or $99 a month

The highest-rated model whose list-price cost for a month of regular use (1,478 requests) stays under each budget. Heavier months cost proportionally more.

Monthly budget	Best model that fits	Est. month
Under $19	Claude Haiku 5.5	$17.73
Under $49	Claude Haiku 5.5	$17.73
Under $99	Claude Sonnet 5.5	$77.42

"Highest-rated" uses our editorial quality and agentic scores; candidates are current models that Claude Code runs with a checked vendor price.

FAQ

Which Claude model is best for Claude Code?

Sonnet 5.5 for everyday work, Opus 5.5 for the hardest tasks, and Haiku 5.5 for cheap subagents. Sonnet 5.5 scores 61.8% on Terminal-Bench 4.0 against Opus 5.5's 64.9%, at half the per-token price, so most sessions can stay on Sonnet and escalate when a task is genuinely hard.

Is Opus 5.5 better than Fable 5.1 in Claude Code?

On the public boards, yes: Opus 5.5 scores 64.9% on Terminal-Bench 4.0 versus 57.9% for Fable 5.1, and 58 versus 53 on the Artificial Analysis Intelligence Index, at $4/$20 per million tokens versus $10/$50. Anthropic says Opus 5.5 performs at Fable 5.1's level on most work.

Should I still use Opus 4.8 or Sonnet 4.6?

Both remain available on the API, but their successors are cheaper and stronger: Opus 5.5 costs $4/$20 versus Opus 4.8's $5/$25, and Sonnet 5.5 costs $2/$10 versus Sonnet 4.6's $3/$15.

Which model does Claude Code use by default?

Claude Opus 5.5 on Pro, Max, Team, Enterprise and Anthropic API accounts, and on Amazon Bedrock and Google Cloud. Microsoft Foundry defaults to Sonnet 4.5. Run /model to switch, or set ANTHROPIC_MODEL; an admin can also set an organization default.

Can Claude Code use non-Claude models?

Yes, through any endpoint that speaks Anthropic's Messages API: set ANTHROPIC_BASE_URL, an auth token and ANTHROPIC_MODEL. DeepSeek, Z.ai, Moonshot and MiniMax run such endpoints for their own models, and gateways cover several vendors on one key. Anthropic does not support these setups, and Claude Code is tuned for Claude, so keep a Claude model available for the hardest steps.

How much do 100 Claude Code requests cost?

At vendor list prices, 100 average coding-agent requests cost: Claude Sonnet 5.5 $5.33, Claude Opus 5.5 $10.66, Claude Haiku 5.5 $1.31, Claude Fable 5.1 $23.76. Measured on real coding-agent traffic through the Standard Compute gateway (Sep 13 to Oct 12, 2026), priced with each request's own token counts and cache reads.

What is the best model for Claude Code on a $19, $49 or $99 monthly budget?

For a month of regular use (1,478 requests, the median in our data), the highest-rated model that fits under $19: Claude Haiku 5.5 (about $17.73); under $49: Claude Haiku 5.5 (about $17.73); under $99: Claude Sonnet 5.5 (about $77.42). Heavier months cost proportionally more.

Every model on this page runs on Standard Compute

Standard Compute

Current frontier models Claude Fable 5.1 · GPT-6 Astra