Claude Opus 4.6

AnthropicProprietary

The first adaptive-thinking Opus. Two generations behind the frontier — exactly the config to watch for how yesterday's flagship ages.

Index @ High

78.0

Context

$/M in · out

$5 · $25

Median speed

55 tok/s

Released

Nov 24, 2025

02·Which effort for which task

The sweet spot, per task

Cheapest effort tier that keeps ≥96% of this model's peak quality on the benchmarks behind each task type. Below the floor number, quality is not okay — it just fails quietly.

Everyday writing & Q&A

Low

Chat, drafting, summaries, general questions.

keeps 96.8% of peak−86% cost

Code assist

Medium

Autocomplete, single functions, small fixes.

keeps 97.4% of peak−66% cost

Coding agents

High

Real repos, multi-file changes, end-to-end tickets.

keeps 100.0% of peakpeak tier needed

Math & quantitative

High

Derivations, statistics, financial models.

keeps 100.0% of peakpeak tier needed

⚠ at Low: only 78% of peak — avoid

Hard research problems

High

Novel, frontier, genuinely difficult questions.

keeps 100.0% of peakpeak tier needed

Agents & tool use

Medium

Long workflows with tools, policies, and state.

keeps 96.9% of peak−66% cost

03·The effort ladder

Quality and cost, tier by tier

On everyday tasks, Low keeps 96.8% of High quality at $0.035 vs $0.24 per task.

Intelligence Index by effort

Cost per typical task by effort

04·Benchmark detail

Scores at Medium

Low Medium High

at Mediumat High (peak)

Benchmark	Medium	Peak	What it predicts
MMLU-Pro	85.0	86.0	Everyday Q&A · Drafting emails & docs · Summarizing articles
GPQA Diamond	80.2	85.0	Technical analysis · Root-cause debugging · Scientific writing
Humanity's Last Exam	23.6	25.0	Novel research questions · Frontier problem solving
AIME 2025	80.6	90.0	Financial modeling · Statistics · Algorithm design
SWE-bench Verified	72.5	76.0	Coding agents · Bug fixing in real repos · Multi-file refactors
LiveCodeBench	76.3	80.0	Writing functions from scratch · Algorithms · Code autocomplete
τ²-bench	78.3	83.0	Tool-using agents · Customer-facing automation · Multi-step workflows
IFEval	90.3	91.0	Structured output (JSON) · Templated generation · Batch pipelines

06·Drift

Is it still the same model this week?

Weekly Intelligence Index · dashed rules mark serving-change events

No serving-change events detected for Claude Opus 4.6 in the tracked window.

07·Call it

One key, this model, your effort

The gateway speaks the OpenAI Chat Completions dialect. Pick a host with claude-opus-4.6@host syntax, set reasoning_effort, and the sweet spot above becomes one line of config. Full docs →

chat with Claude Opus 4.6

curl https://models.fyi/api/v1/chat/completions \
  -H "Authorization: Bearer $MFYI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "claude-opus-4.6",
    "reasoning_effort": "low",
    "messages": [{"role": "user", "content": "Hello"}]
  }'
# local dev: replace https://models.fyi with http://localhost:3000