Claude Fable 5

AnthropicProprietary

Anthropic's Mythos-class flagship, above Opus in capability. Current leader on real-repo coding and agentic work.

Index @ High

82.2

Context

$/M in · out

$10 · $50

Median speed

65 tok/s

Released

Jun 2, 2026

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 96.5% of peak−95% cost

Code assist

Medium

Autocomplete, single functions, small fixes.

keeps 96.6% of peak−88% cost

Coding agents

High

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

keeps 98.0% of peak−65% cost

Math & quantitative

High

Derivations, statistics, financial models.

keeps 96.7% of peak−65% cost

⚠ at Low: only 76% of peak — avoid

Hard research problems

High

Novel, frontier, genuinely difficult questions.

keeps 97.5% of peak−65% cost

Agents & tool use

High

Long workflows with tools, policies, and state.

keeps 98.7% of peak−65% cost

03·The effort ladder

Quality and cost, tier by tier

On everyday tasks, Low keeps 96.5% of X-High quality at $0.080 vs $1.54 per task.

Intelligence Index by effort

Cost per typical task by effort

04·Benchmark detail

Scores at X-High

Low Medium High X-High

at X-Highat X-High (peak)

Benchmark	X-High	Peak	What it predicts
MMLU-Pro	90.0	90.0	Everyday Q&A · Drafting emails & docs · Summarizing articles
GPQA Diamond	90.0	90.0	Technical analysis · Root-cause debugging · Scientific writing
Humanity's Last Exam	37.0	37.0	Novel research questions · Frontier problem solving
AIME 2025	96.0	96.0	Financial modeling · Statistics · Algorithm design
SWE-bench Verified	82.0	82.0	Coding agents · Bug fixing in real repos · Multi-file refactors
LiveCodeBench	86.0	86.0	Writing functions from scratch · Algorithms · Code autocomplete
τ²-bench	88.0	88.0	Tool-using agents · Customer-facing automation · Multi-step workflows
IFEval	94.0	94.0	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 Fable 5 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-fable-5@host syntax, set reasoning_effort, and the sweet spot above becomes one line of config. Full docs →

chat with Claude Fable 5

curl https://models.fyi/api/v1/chat/completions \
  -H "Authorization: Bearer $MFYI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "claude-fable-5",
    "reasoning_effort": "low",
    "messages": [{"role": "user", "content": "Hello"}]
  }'
# local dev: replace https://models.fyi with http://localhost:3000