Claude 3.5 Haiku vs K2-V2 (high)
Wins 0 of 5 areas
—
Wins 4 of 5 areas
Coding · Reasoning · Long documents · Following instructions
K2-V2 (high) is the stronger all-rounder.
Scores updated · 12 tests both models report · How we compare
Where each one wins
Tests won in each of the five areas where both have results. Each piece is one test, so a longer bar means more evidence; grey means the two scored within a point of each other.
- ReasoningHard problems that need careful thinking02K2-V2 (high)2 of 3 tests · 1 tie
- CodingWriting and fixing software02K2-V2 (high)2 of 2 tests
- Long documentsFinding answers in very long texts01K2-V2 (high)1 of 1 test
- Following instructionsDoing exactly what it is asked01K2-V2 (high)1 of 1 test
- FactsGetting facts right instead of making them up11Even1 each
Long documents and following instructions rest on a single test each.
The biggest differences
The tests each model wins by the widest margin, up to three each. Scores are out of 100.
Where K2-V2 (high) pulls ahead
- Graduate-level biology, physics and chemistry questionsGPQA Diamond+27.3points ahead
- Follows unfamiliar, precisely checkable instructionsIFBench+17.3points ahead
- Reasons across sets of long documentsAA-LCR+7.7points ahead
Where Claude 3.5 Haiku pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+49.9points ahead
Every test, side by side
All 12 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingK2-V2 (high)
- Terminal-Bench HardK2-V2 (high) by 7.52.39.8+7.5
- SciCodeK2-V2 (high) by 1.227.428.6+1.2
ReasoningK2-V2 (high)
- GPQA DiamondK2-V2 (high) by 27.340.868.1+27.3
- Humanity's Last ExamK2-V2 (high) by 6.93.610.5+6.9
- CritPttie00tie
FactsEven
- AA-Omniscience · Non-hallucinationClaude 3.5 Haiku by 49.958.99+49.9
- AA-Omniscience · AccuracyK2-V2 (high) by 5.313.218.5+5.3
Following instructionsK2-V2 (high)
- IFBenchK2-V2 (high) by 17.342.860.1+17.3
Other results3 tests, not counted
Tests outside the eight areas. They are not counted above: several are summary scores built from other tests, or the same test under another name.
- AA-OmniscienceClaude 3.5 Haiku by 33.1-22.5-55.6+33.1
- τ²-Bench Telecom (AA run)K2-V2 (high) by 3.224.627.8+3.2
- AA IntelligenceK2-V2 (high) by 18.99.9+1
Questions people ask
Which is better, Claude 3.5 Haiku or K2-V2 (high)?
K2-V2 (high) wins four of the five areas where both have results: coding, reasoning, long documents and following instructions. Claude 3.5 Haiku wins none. They are level on facts.
Which is better for coding?
K2-V2 (high). It wins 2 of the 2 coding tests both models report; Claude 3.5 Haiku wins none.
How do you compare the two?
We use the 12 benchmark tests both models have published scores on. The verdict counts the 9 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 3 are listed but not counted, because several are summary scores or repeat a test. Each score is the one shown on the model's own page.