Claude Sonnet 4.6 vs Claude Fable 5: Benchmarks, Pricing & Capabilities Compared
TL;DR — Claude Sonnet 4.6 wins for cost · Claude Fable 5 wins for reasoning + long-context.
Claude Sonnet 4.6 Anthropic
- Released
- 2026-02-17
- Context window
- 500K tokens
- Input price
- $3.00 / Mtok
- Output price
- $15.00 / Mtok
Key features
- Agent Teams: orchestrate 2-16 Claude instances
- Near-Opus performance at 1/5th cost
- 80.8% SWE-bench Verified
Claude Fable 5 Anthropic
- Released
- 2026-06-09
- Context window
- 1M tokens
- Input price
- $10.00 / Mtok
- Output price
- $50.00 / Mtok
Benchmark comparison
| Benchmark | Claude Sonnet 4.6 | Claude Fable 5 |
|---|---|---|
| GPQA Diamond | 79.7% | 92.6% ✓ |
| HLE | 10.8% | 53.3% ✓ |
| IF-Bench | 42.4% | 63.5% ✓ |
| LiveCodeBench Reasoning | 58.7% | 70.0% ✓ |
| SciCode | 44.1% | 60.2% ✓ |
| TAU2-bench | 78.9% | 98.5% ✓ |
| TerminalBench-Hard | 42.4% | 62.9% ✓ |
Pricing comparison
| Metric | Claude Sonnet 4.6 | Claude Fable 5 |
|---|---|---|
| Input ($/Mtok) | $3.00 | $10.00 |
| Output ($/Mtok) | $15.00 | $50.00 |
| Cached input ($/Mtok) | $0.30 | — |
| Cost per 1M-token roundtrip (1M in + 1M out) | $18.00 | $60.00 |
Context window & modalities
| Attribute | Claude Sonnet 4.6 | Claude Fable 5 |
|---|---|---|
| Context window | 500K tokens | 1M tokens |
| Input modalities | text, image, PDF | text, image |
| Output modalities | text | text |
| Knowledge cutoff | 2025-10 | — |
Verdict by use case
Coding
Insufficient data
Basis: SWE-bench
No shared coding benchmark.
Reasoning
→ Claude Fable 5
Basis: GPQA Diamond
Claude Sonnet 4.6 79.7% vs Claude Fable 5 92.6% on GPQA Diamond.
Math
Insufficient data
Basis: MATH / AIME
No shared math benchmark.
Long context
→ Claude Fable 5
Basis: Context window
Claude Sonnet 4.6 500K tokens vs Claude Fable 5 1M tokens.
Cost
→ Claude Sonnet 4.6
Basis: Input $/Mtok
Claude Sonnet 4.6 $3/Mtok vs Claude Fable 5 $10/Mtok input.
Changelog & releases
Claude Sonnet 4.6
Released 2026-02-17
Predecessor: anthropic-claude-sonnet-4
- Agent Teams: orchestrate 2–16 Claude instances in parallel
- +8.5pt on SWE-bench Verified vs Sonnet 4
- 1/5 the cost of Opus 4.5 at ~95% of coding quality
- Fast mode research preview for lower-latency inference
Claude Fable 5
Released 2026-06-09