Side-by-side comparison · models
Claude 3 Haiku vs Claude Fable 5.1
What we have checked about each, line by line. Every value shows where it came from and when we last confirmed it.
Models
Claude 3 Haiku
Claude 3 Haiku is Anthropic's fastest and most compact model for near-instant responsiveness. Quick and accurate targeted performance.
Models
Claude Fable 5.1
Claude Fable 5.1 improves on Claude Fable 5 across the board, with the biggest gains in agentic coding, long-running agentic workflows, and knowledge work: long code refactors.
Checked facts
Line by line
| Measure | Claude 3 Haiku | Claude Fable 5.1 |
|---|---|---|
| Input price | Not checked yet | $10.00 per million tokens |
| Output price | Not checked yet | $50.00 per million tokens |
| Price on amazon-bedrock | Not checked yet | $0.250 per million cached input tokens |
| Price on amazon-bedrock | Not checked yet | $10.00 per million input tokens |
| Price on amazon-bedrock | Not checked yet | $50.00 per million output tokens |
| Price on anthropic | Not checked yet | $0.250 per million cached input tokens |
| Price on anthropic | Not checked yet | $10.00 per million input tokens |
| Price on anthropic | Not checked yet | $50.00 per million output tokens |
| Price on azure | Not checked yet | $0.250 per million cached input tokens |
| Price on azure | Not checked yet | $10.00 per million input tokens |
| Price on azure | Not checked yet | $50.00 per million output tokens |
| Price on google-vertex | Not checked yet | $0.250 per million cached input tokens |
| Price on google-vertex | Not checked yet | $10.00 per million input tokens |
| Price on google-vertex | Not checked yet | $50.00 per million output tokens |
| Latest version | Not checked yet | anthropic/claude-fable-5.1-20260831 |
| Time to first token (ms) | Not checked yet | 97,890 |
| Time to first token (ms) on amazon-bedrock | Not checked yet | 2,912 |
| Time to first token (ms) on anthropic | Not checked yet | 2,759 |
| Time to first token (ms) on azure | Not checked yet | 4,508 |
| Time to first token (ms) on google-vertex | Not checked yet | 2,529 |
| Output speed (tokens a second) | Not checked yet | 61.3 |
| Output speed (tokens a second) on amazon-bedrock | Not checked yet | 54 |
| Output speed (tokens a second) on anthropic | Not checked yet | 63 |
| Output speed (tokens a second) on azure | Not checked yet | 59 |
| Output speed (tokens a second) on google-vertex | Not checked yet | 69 |
| Tokens served on OpenRouter in a day | 3.9M (for 26 Sept 2026) | 25.9B (for 7 Nov 2026) |
| Tool calls on OpenRouter in a day | 81 (for 26 Sept 2026) | 134.9K (for 7 Nov 2026) |
| Artificial Analysis Intelligence Index | 5.6 | 50 |
| LMArena text rating (Elo) | 1,195 | 1,513 |
| LMArena vision rating (Elo) | 950 | 1,321 |
| Gpqa diamond | 36.3% (for 7 Mar 2024) | Not checked yet |
| Cadeval | 12% (for 7 Mar 2024) | Not checked yet |
| Dtbench | 50.1% (for 7 Mar 2024) | 97.6% (for 1 Sept 2026) |
| Lmca | 8.8% (for 7 Mar 2024) | 65.5% (for 1 Sept 2026) |
| Math level 5 | 14.9% (for 7 Mar 2024) | Not checked yet |
| Mmlu | 73.8% (for 7 Mar 2024) | Not checked yet |
| Weirdml | 9.8% (for 7 Mar 2024) | 92.9% (for 1 Sept 2026) |
| Winogrande | 74.2% (for 7 Mar 2024) | Not checked yet |
| Otis mock aime 2024 2025 | 1.8% (saturated benchmark: the top models all score near the ceiling) (for 7 Mar 2024) | 100% (saturated benchmark: the top models all score near the ceiling) (for 1 Sept 2026) |
| Artificial Analysis Coding Index | Not checked yet | 81.6 |
| Artificial Analysis Agentic Index | Not checked yet | 57.9 |
| LMArena web development rating (Elo) | Not checked yet | 1,746 |
| Design Arena rating (Elo) | Not checked yet | 1,424 |
| Frontiermath tiers 1 3 v2 private | Not checked yet | 90.2% (for 1 Sept 2026) |
| Simpleqa verified | Not checked yet | 70.8% (for 1 Sept 2026) |
| Arc agi 2 | Not checked yet | 90% (for 1 Sept 2026) |
| Apex agents | Not checked yet | 68.6% (for 1 Sept 2026) |
| Arc agi | Not checked yet | 97.5% (for 1 Sept 2026) |
| Chess puzzles | Not checked yet | 48% (for 1 Sept 2026) |
| Ebr bench | Not checked yet | 57.1% (for 1 Sept 2026) |
| Frontiercode | Not checked yet | 50.9% (for 1 Sept 2026) |
| Frontiermath tier 4 v2 private | Not checked yet | 87.8% (for 1 Sept 2026) |
| Frontierswe | Not checked yet | 56.3% (for 1 Sept 2026) |
| Furniture assembly | Not checked yet | 70% (for 1 Sept 2026) |
| Gso bench | Not checked yet | 88.2% (for 1 Sept 2026) |
| Mirrorcode | Not checked yet | 73.3% (for 1 Sept 2026) |
| Mystery game puzzles | Not checked yet | 58% (for 1 Sept 2026) |
| Proofbench | Not checked yet | 100% (for 1 Sept 2026) |
| Remote labor index | Not checked yet | 17.9% (for 1 Sept 2026) |
| Spireclimb | Not checked yet | 2% (for 1 Sept 2026) |
| Weirdml v3 | Not checked yet | 43.5% (for 1 Sept 2026) |
A line appears only when at least one record has a value we confirmed from a source. Values we have not confirmed are left out, never guessed. Records of different types share fewer lines.
Add a third
Compare with one more record
How this works
Share it, or build your own
This page works for any two or three records in the directory, of any type. The link stays the same whichever order you name them in, so you can share it as it is. Change the records in the address, for example /directories/vs/claude-code-vs-cursor, or start from our editorial comparisons.