Claude Fable 5 vs Hy3 preview

Price, context and benchmark scores side by side — every number from the live model catalogue.

Claude Fable 5anthropic Hy3 previewtencent
Input / 1M tokens $10.00 $0.063
Output / 1M tokens $50.00 $0.210
1M in + 300K out $25.00 $0.13
Context window 1M 262K
Intelligence index 59.9 41.2
Coding index 76.5 58.8
Agentic index 52.8 30.7
Cheapest provider Azure GMICloud
Providers serving it 4 1
Weights Closed Open
Image input Yes No
Tool calling Yes Yes
Reasoning mode Yes Yes
Released Jun 9, 2026 Apr 22, 2026

Which one to pick

Pick Claude Fable 5 if…

  • You want the higher general-intelligence score (59.9).
  • Code quality is the priority — it leads on the coding index (76.5).
  • You are running agent loops — higher agentic index (52.8).
  • You need the bigger context window (1M).
  • You want provider redundancy: 4 providers serve it.

Pick Hy3 preview if…

  • Token cost matters: $0.210 per 1M output vs $50.00.
  • You want the option to self-host — weights are public.

FAQ

Is Claude Fable 5 better than Hy3 preview?

Claude Fable 5 scores higher on general intelligence (59.9 vs 41.2), while Hy3 preview is 238× cheaper per output token ($0.210 vs $50.00). For code, Claude Fable 5 leads (76.5 coding index). Claude Fable 5 takes the larger context at 1M.

Which is cheaper, Claude Fable 5 or Hy3 preview?

Hy3 preview, at $0.210 per 1M output tokens versus $50.00. On a 1M input + 300K output workload that is $0.13 against $25.00.

Can I self-host Claude Fable 5 or Hy3 preview?

Claude Fable 5: closed weights, API only. Hy3 preview: open weights (tencent/Hy3-preview).

Related comparisons