Claude Fable 5 vs Claude Haiku 5.5
Claude Fable 5 leads Claude Haiku 5.5 by 6.6 points on the Shogo Agent Index, but their ranges overlap, so treat it as close. Claude Fable 5 wins 18 and Claude Haiku 5.5 wins 2 of 20 shared benchmarks. Claude Haiku 5.5 is cheaper for agent workloads.
Add more models to this comparison →Pillars
| Pillar | Claude Fable 5 | Claude Haiku 5.5 |
|---|---|---|
| Agentic coding | 74 | 67 |
| Long-horizon autonomy | 69 | 62 |
| Knowledge-work agents | 52 | 46 |
| Computer use | 55 | 47 |
| Security & context learning | 59 | 55 |
Shared benchmarks
| Benchmark | Claude Fable 5 | Claude Haiku 5.5 |
|---|---|---|
| APEX-Agentsindex | 63.6% | 56.1% |
| Vending-Bench 2index | $5,680 | $3,602 |
| FrontierSWEindex | 47.0% | 43.8% |
| ARC-AGI | 98.5% | 89.0% |
| ARC-AGI-2 | 89.2% | 61.3% |
| Chess Puzzles | 47.0% | 55.0% |
| CritPt | 28.6% | 22.6% |
| DTBench | 98.4% | 95.0% |
| FrontierCode | 53.5% | 46.4% |
| FrontierMath Tier 4 | 90.2% | 46.3% |
| FrontierMath Tiers 1-3 | 87.0% | 75.1% |
| GPQA Diamond | 93.9% | 89.6% |
| LMCA | 61.2% | 51.1% |
| MirrorCode | 70.0% | 78.9% |
| OTIS Mock AIME | 100.0% | 98.9% |
| ProofBench | 95.0% | 89.0% |
| SciCode | 61.0% | 55.0% |
| SimpleBench | 81.9% | 59.2% |
| SimpleQA Verified | 70.7% | 23.8% |
| WeirdML v3 | 15.6% | 11.7% |
Specs and price
| Claude Fable 5 | Claude Haiku 5.5 | |
|---|---|---|
| Agent Index | 64.1 (#12) | 57.5 (#18) |
| Agent $/M | $4.62 | $0.050 |
| Input / output $/M | $10.00 / $50.00 | $0.10 / $0.50 |
| Context | 1M | 1M |
| Released | Jun 9, 2026 | Oct 7, 2026 |
| Weights | Closed | Closed |
| API hosts | 4 | 5 |
Don’t pick one. Route.
Shogo can run Claude Fable 5 and Claude Haiku 5.5 side by side and send each step of an agent to the one that’s cheaper for it.
Try both in Shogo