Breaking

Anthropic Launches Claude Fable 5.1, Outscoring GPT-5.6 Sol on Agentic Tests

September 1, 2026 at 18:21 EDT

Foundation Models
Software Dev & Coding
AI Agents

Anthropic on September 1, 2026 released Claude Fable 5.1, a generally available Mythos-class model that immediately became the new comparison point against OpenAI’s GPT-5.6 Sol. Independent evaluators placed Fable 5.1 at the top of aggregate intelligence scores, while same-day one-shot coding trials showed Sol remaining far cheaper for comparable visual output.

September 1, 2026 · Anthropic

Fable 5.1 vs GPT-5.6 Sol

Anthropic’s new flagship leads independent intelligence scores and vendor agentic tests. A same-day one-shot coding trial still bills about six times as much as OpenAI’s Sol.

Key figures

Fable 5.1 AA Index (max)

GPT-5.6 Sol AA Index (max)

75%

Cache-read price cut

6×

One-shot Three.js cost gap

Magnitude · one-shot coding cost

$5.69 vs $0.88 on the same five-scene test

$0.88

GPT-5.6 Sol

$5.69

Claude Fable 5.1

Six stacked blocks versus one: about 6× the bill for denser Three.js geometry, with bugs on both sides.

Comparison · Terminal-Bench-Science 0.1

Agentic science accuracy, drawn to scale

22.4%

GPT-5.6 Sol

52.6%

Claude Fable 5.1

Column height tracks the score: Fable 5.1 more than doubles Sol on this Anthropic-reported science-agent bench.

Where Fable 5.1 lands ahead

Highest AA Intelligence Index (66). Stronger vendor agentic scores. Enterprise testers cite long unattended coding runs and a rare crash Fable 5 missed.

Where the gap still hurts

List price remains $10/$50 versus Sol’s $5/$30. Max-effort runs cost ~20% more than Fable 5. Vals still has Sol slightly ahead on Terminal-Bench 2.1. Usage windows burn fast.

Pricing catch

Cache reads fell to $0.25 per million tokens, but max-effort output tokens rose about 1.7×.

Typical claim: ~25% cheaper Agentic claim: up to ~45% Max effort measured: +20% vs Fable 5

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.