Skip to content
fedi.software

Claude Haiku 5.5 in fedi-index

The score in every slice, what it is made of, value, local run and history.

fedi-index · General

A 1,430 ±6

Percentile 80 · #60

People prefer / Solves tasks

1,396 / 1,481

Human votes vs benchmarks, Elo-eq

Value

+105◆

$0.20 per 1M tokens, median of 44 providers · vendor $0.20

On your hardware

Closed weights: cloud only.

Score decomposition

A General 1,430 [1,423–1,436] Percentile 80 · #60 · components: 9
Source Component Value Elo-eq Weight Measured Variant
Epoch AI eci 160 ±1.17 1,520 ±9 33.2% 2026-10-07 —
LMArena text/overall 1,398 ±3.14 1,398 ±3 16.8% 2026-11-11 high
LMArena text/hard_prompts 1,418 ±3.77 1,412 ±3 16.6% 2026-11-11 high
LMArena text/instruction_following 1,411 ±4.80 1,424 ±5 8.0% 2026-11-11 high
LMArena text/expert 1,443 ±8.54 1,429 ±7 7.2% 2026-11-11 high
LMArena text/longer_query 1,395 ±4.47 1,396 ±4 4.1% 2026-11-11 high
LMArena text/multi_turn 1,388 ±7.07 1,388 ±6 3.7% 2026-11-11 high
Epoch AI simpleqa_verified outlier decay ×0.58 0.24 ±0.01 1,372 ±4 2.1% 2026-10-09 max
Epoch AI gpqa_diamond outlier decay ×0.25 0.90 ±0.02 1,457 ±9 0.7% 2026-10-09 max

Weight is the share of the component in the score (the rest is the prior that pulls models with little data towards the population median). Methodology

A Coding 1,436 [1,433–1,440] Percentile 86 · #40 · components: 3
Source Component Value Elo-eq Weight Measured Variant
LMArena webdev/overall 1,593 ±4.20 1,476 ±1 35.2% 2026-11-15 high
LMArena text/coding 1,454 ±5.68 1,429 ±5 31.8% 2026-11-11 high
Terminal-Bench tbench/terminal-bench decay ×0.90 0.40 ±0.02 1,474 ±4 24.2% 2026-10-07 max

Weight is the share of the component in the score (the rest is the prior that pulls models with little data towards the population median). Methodology

B Agents 1,437 [1,433–1,440] Percentile 66 · #27 · components: 2
Source Component Value Elo-eq Weight Measured Variant
LMArena agent/overall -0.00 ±0.01 1,454 ±2 67.6% 2026-11-15 high
Terminal-Bench tbench/terminal-bench decay ×0.90 0.40 ±0.02 1,474 ±4 23.7% 2026-10-07 max

Weight is the share of the component in the score (the rest is the prior that pulls models with little data towards the population median). Methodology

A Math 1,431 [1,420–1,442] Percentile 84 · #43 · components: 3
Source Component Value Elo-eq Weight Measured Variant
LMArena text/math 1,436 ±11 1,427 ±11 46.1% 2026-11-11 high
Epoch AI frontiermath_t1_3 decay ×0.82 0.75 ±0.03 1,484 ±6 33.8% 2026-10-07 max
Epoch AI frontiermath_t4 decay ×0.82 0.46 ±0.08 1,466 ±13 11.3% 2026-10-07 max

Weight is the share of the component in the score (the rest is the prior that pulls models with little data towards the population median). Methodology

C Writing 1,364 [1,356–1,371] Percentile 50 · #133 · components: 2 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/creative_writing 1,337 ±6.02 1,364 ±6 60.5% 2026-11-11 high
LMArena text/longer_query 1,395 ±4.47 1,396 ±4 32.8% 2026-11-11 high

Weight is the share of the component in the score (the rest is the prior that pulls models with little data towards the population median). Methodology

C Language: 繁體中文 1,386 [1,373–1,400] Percentile 48 · #114 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/chinese 1,433 ±9.33 1,400 ±7 92.8% 2026-11-11 high

Weight is the share of the component in the score (the rest is the prior that pulls models with little data towards the population median). Methodology

B Language: English 1,382 [1,373–1,390] Percentile 54 · #123 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/english 1,406 ±4.61 1,394 ±5 93.5% 2026-11-11 high

Weight is the share of the component in the score (the rest is the prior that pulls models with little data towards the population median). Methodology

C Language: Español 1,378 [1,356–1,400] Percentile 28 · #109 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/spanish 1,397 ±13 1,396 ±12 90.5% 2026-11-11 high

Weight is the share of the component in the score (the rest is the prior that pulls models with little data towards the population median). Methodology

C Language: Français 1,391 [1,368–1,414] Percentile 38 · #89 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/french 1,427 ±14 1,411 ±13 90.3% 2026-11-11 high

Weight is the share of the component in the score (the rest is the prior that pulls models with little data towards the population median). Methodology

C Language: Русский 1,373 [1,362–1,385] Percentile 45 · #126 · components: 1 provisional
Source Component Value Elo-eq Weight Measured Variant
LMArena text/russian 1,377 ±7.24 1,386 ±6 93.1% 2026-11-11 high

Weight is the share of the component in the score (the rest is the prior that pulls models with little data towards the population median). Methodology

History

fedi-index by month
The same in a table
Model 2026-11
Claude Haiku 5.5 1,430 A

Badge

Show the model's fedi-index on your site or in a README. The badge updates itself; it links to this page.

fedi-index: Claude Haiku 5.5

Data CC BY 4.0: keep the link to fedi.software.